United States
Environmental Protection
Agency
Office of
Research and
Development
Office of Solid Waste
and Emergency
Response
EPA/540/S-97/505
January 1998
                         Ground  Water   Issue
                          Steam Injection for Soil  and  Aquifer Remediation
                          Eva L. Davis
Background
  Innovative technologies for subsurface remediation, including
in situ techniques based on heating the subsurface to enhance
the  recovery  of organic contaminants, are increasingly being
evaluated for use at specific sites as the  limitations to the
conventionally-used techniques are recognized.  The purpose
of this Issue Paper is to provide to those involved in assessing
remediation technologies for specific sites basic technical
information on the use of steam injection for the remediation of
soils and aquifers that are contaminated by volatile or semivolatile
organic compounds.  A related Issue Paper, entitled "How Heat
Can Enhance In-Situ Soil and Aquifer Remediation: Important
Chemical Properties and Guidance on Choosing the Appropriate
Technique" (Davis, 1997), discusses the properties of some
organic chemicals commonly found at contaminated sites, how
these properties are affected by the presence of the chemical in
a porous media, and how heat can enhance the recovery of these
chemicals from the subsurface. The Issue Paper also provides
information on three general types of heat-based remediation
systems,  and some guidance on which technique  is  most
appropriate for different  soil and  aquifer conditions and for
different organic chemicals.
  This document  contains more detailed information on how
steam injection can be used to recover organic contaminants
from the subsurface, the contaminant and subsurface conditions
for which the  process may be appropriate, and general design
and equipment considerations. In addition, laboratory and field-
scale experiments are described, and available treatment cost
  U.S. Environmental Protection Agency
    National Risk Management Research Laboratory
    Subsurface Protection and Remediation Division
    P. O. Box 1198, Ada, OK 74820
information is provided. This document is not meant to provide
detailed  information that would allow the design of a steam
injection remediation project, but rather design considerations
are provided to familiarize  remediation workers with what is
involved in the process.
Introduction
  Steam injection was first developed by the petroleum industry
for the enhanced recovery of oils from reservoirs. In petroleum
industry applications, steam is injected to lower the viscosity of
heavy oils and to increase the volatility of light oils. As much as
50 percent of the original oil in place may remain in the reservoir
when the process becomes  uneconomical and is discontinued.
In the past several years, steam injection has been adapted for
the recovery of organic contaminants from the subsurface, and
extensive laboratory and field research has been done. When
steam injection is used for subsurface remediation, the objective
is to remove as much of the contamination as possible, thus
reducing the residualto very lowlevels. The subsurface conditions
dealtwith by the petroleum industry versus remediation purposes
are generally very different - the petroleum industry dealing with
deep, confined reservoirs and the remediation industry with the
shallow, generally unconfined subsurface. Thus, the petroleum
industry technique and the technique for remediation purposes
differ in significant ways.
Basic Principles
  Consider the situation shown in  Figure  1  where  steam is
injected  surrounding a pool of a volatile contaminant in the
subsurface.   The figure illustrates steam injection above the
water table,  but steam injection for remediation purposes has
also been successfully carried out belowthe watertable (Newmark
and Aines, 1995),  and the basic principles described here are
the same for either situation. Initially, the steam that is injected
will heat the well bore, and  the formation around the injection
                         Superfund Technology Support Center for
                         Ground Water

                          Robert S. Kerr Environmental
                          Research Laboratory
                          Ada, Oklahoma
                      Technology Innovation Office
                      Office of Solid Waste and Emergency
                      Response, US EPA, Washington, DC

                      Walter W. Kovalick, Jr., Ph.D.
                      Director

-------
      _t           Water, Vapor, NAPL
      Steam                               Steam
           r////////////////
Figure 1. Diagram of a possible steam injection remediation system.

zone of the well.  The steam condenses as the latent heat of
vaporization of water is transferred from the steam to the well
bore and the porous media where it enters the formation. As
more steam is injected, the hot water moves into the formation,
pushing the water initially in the formation (which is at ambient
temperature) further into the porous media. When the porous
media at the point of steam injection has absorbed enough heat
to reach  the temperature of the  injected  steam, steam itself
actually enters the media, pushing the cold water and the bank
of condensed steam (hot water) in front of it.
  As these flowing fluids approach a  region that contains the
volatile contaminant  at saturations greater than  its  residual
saturation, the contaminant is displaced.  First to come into
contact with the contaminant is cold water, then the hot water
bank, and finally the steam front.  The cold water will flush the
mobile  contaminant (i.e., the contaminant saturation that is  in
excess of its residual saturation) from the pores. The hot water
will reduce the viscosity of the contaminant, making it easier to
be displaced by viscous forces, and  may reduce the residual
saturation of the contaminant. When the steam front reaches the
contaminated area, no additional contaminant can be recovered
by  viscous forces.   Additional  recovery  is  achieved by
volatilization, evaporation, and/or steam distillation of the volatile
and semivolatile contaminants (Stewart and Udell, 1988).
  Thus, when steam is injected  into a porous media, three
distinct zones can be considered to develop: the steam zone,
the variable temperature zone, and the ambient temperature
zone. These three temperature zones are shown in Figure 2.
This figure  also  shows the  relative concentrations of the
contaminants in each of the zones. Closest to the injection point
is the steam zone which, in the area close around the injection
point, is at approximately the temperature of the steam.  Further
downstream within  the steam zone, the temperature may
decrease somewhat due to heat losses to the overburden and
underburden. If the rate of steam injection is sufficiently high, the
percentage of heat lost will not be great,  and  this zone can be
considered essentially isothermal.  Inthiszone, steam distillation
and steam stripping are the  main recovery mechanisms and
these processes cause the formation of a contaminant bank just
downstream of the steam front (Wu, 1977). Downstream of this
is  a variable temperature  zone where condensation of the
injected steam and evaporation of the contaminants takes place.
The interface between the steam zone  and the variable
temperature zone is essentially  the location  of steam
condensation, if the temperature gradient in the steam zone is
not significant. Atthe front ofthiszone is the bankof contaminants
that have been displaced by viscous forces, followed by the hot
water bank.  The  zone furthest downstream is  the  ambient
temperature zone which is saturated with water and the mobilized
contaminant.
  The amount of residual saturation remaining after the cold and
hot water flush is dependent on the capillary properties of the
porous media, the interfacial properties of the contaminant, and
the pressure gradient  in the water causing the displacement.
Residual organic liquid saturations in unconsolidated sands are
typically in the range of 14  to 30 percent (Wilson et al., 1990).
The texture of the soil has been found to be the most important
factor in determining the amount of residual organic contaminant
left in a soil. Several researchers have found that increasing the
temperature reduces the residual saturation of oils (Edmondson,
1965; Poston etal., 1970; Davis and  Lien, 1993).
  The residual contaminants are volatilized by the steam, and
the volatilized contaminants are transported to the steam front,
increasing the saturation of the contaminant in this zone and
adding to the contaminant bank. The rate atwhich the contaminant
bank formed by the evaporation and condensation processes
moves downstream is inversely proportional to the saturation of
the compound downstream, and directly proportional to the
volatility of the contaminant (Yuan, 1990). Experimental work
has shown that pure, separate,  liquid-phase contaminants with
boiling points less than that of water will be completely removed
from the steam zone  except for the small amount which is
adsorbed to solid surfaces or dissolved in liquid water which may
be present  in the steam zone (Hunt et al., 1988).  Theoretical
studies  have predicted that under certain conditions, liquid
hydrocarbons having  boiling  points  up to 175°C may also  be
completely  removed directly behind  the steam condensation
front (Yuan, 1990; Falta et  al., 1992b).  This  conclusion is
supported  by one-dimensional column experiments  which
                         Hot water
                           bank
        Contaminant bank
        containing volatile
        components
Contaminants
displaced by
viscous forces
                                               Zones
                                               Ambient
                                               temperature
                                               zone
 o
 O
              Distance from injection well


Figure 2. The three temperature zones that form during a steam flood
(after Wu, 1977).

-------
showed essentially complete removal of toluene and gasoline
(Hunt et al., 1988) and 96.8 to 99.8 percent recovery of No. 2 fuel
oil and jet fuel (Hadim et al., 1993) by steam injection.

Recovery Mechanisms
  The first recovery mechanism acting on contaminants during
a steam injection is a physical displacement, first by the water
originally in the subsurface, then by the hot water formed by the
condensation of steam and finally by the steam itself. Physical
displacement of the  contaminant occurs  when  there  is an
immiscible organic liquid present at saturations greater than its
residual saturation, and when there is a dissolved phase in the
water that is displaced with the water. A physical displacement
of soil air occurs when  steam is injected into  the unsaturated
zone. When the initial contaminant concentration is significantly
greater than residual saturation and its volatility is low, the
greatest reduction in its saturation will be due to  displacement by
water (Herbeck et al., 1976).
  The  increased  temperatures  that accompany the  steam
injection  process will cause decreases in the  capillary and
interfacial forces between fluids and the porous  media which will
reduce the  residual saturation of the organic phase behind the
hot water front.  Also, thermal expansion of the organic phase
can increase its saturation,  increasing the  mobile fraction and
decreasing the residual.   Another  effect of the  increased
temperature is a decrease in the viscosity of the organic phase,
which also increases its mobility.  All of these factors contribute
to the formation of a bank  of the organic phase in front of the
steam front that is displaced by physical forces.
  The main recovery mechanisms for contaminants in the steam
zone are steam distillation (also called co-distillation) and steam
displacement (stripping). Steam distillation occurs when a liquid
that is immiscible with water is present.  The boiling point of the
mixture is reached when the total vapor pressure of the system
becomes equal to one atmosphere, rather than when the vapor
pressure of the individual  component becomes equal to one
atmosphere.  Since both liquids contribute to the total vapor
pressure, this point is reached at a lower temperature than the
normal boiling point of either of the liquids alone (Atkins, 1986).
Thus, some immiscible contaminants that have a normal boiling
point that is greater than 100°C may also be readily removed by
the steam injection process.  The steam distillation  process is
dependent  on the composition of the  liquids, as well  as the
temperature and pressure of the system. Steam stripping occurs
as the injected  steam sweeps the contaminant vapor to the
condensation front, where the vapors condense, increasing the
saturation of the liquid contaminant at  the condensation  front
and allowing additional  vaporization of the contaminant  in the
steam zone. When the saturation  of the contaminant exceeds its
solubility at ambient temperature, a contaminant bank is formed
in front of the steam zone.
  When the contaminant is a mixture of volatile  and semivolatile
components, such  as gasoline or kerosene, the lower boiling
components will vaporize first due to their largervapor pressures.
As a result of the removal of the volatile components, the liquid
phase  concentrations and  thus the vapor pressures  of the
remaining "semi" and "nonvolatile" components  in the  liquid
phase will increase, which leads to an increase in the evaporation
rates of these components in the steam zone (Yuan, 1990). This
sequential vaporization  of  compounds means that the
contaminant  bank which  is  recovered about  the time of
breakthrough of the steam  will be enriched in the  lighter, more
volatile components of the contaminant (Weyland  etal., 1991).
By continuing steam injection after steam breakthrough, the less
volatile components can be recovered at greater rates as their
concentration in  the remaining liquid, and  thus in the  vapor
phase, increases. Udell and  McCarter (1996) showed in one-
dimensional laboratory experiments that compounds with boiling
points up to approximately 300°C were removed by continuing
injection of steam past breakthrough while the concentrations of
compounds with boiling points up to 450°C were reduced by at
least an order of magnitude by the injection of 100 pore volumes
of steam. However, asthe volume of remaining liquid decreases
and it recedes into smaller pores, interfacial effects increase and
the vapor pressure of the contaminants decreases, and this
ultimately limits the amount of contaminants that can be recovered.
The  higher the temperature, the greater the vapor pressure
(within the limits of the interfacial effects) and thus the greaterthe
recovery (Lingineni and Dhir,  1992).
  Cycling  of steam injection and vacuum extraction after steam
breakthrough at the extraction well has been found to be very
effective  during  field demonstrations  for the  recovery  of
contaminants (Udell and  Stewart, 1989; Newmark and Aines,
1995). Itamura and Udell (1995) have shown theoretically and
numerically that depressurizing the steam zone by halting steam
injection  while continuing   vapor  extraction will  cause a
thermodynamically unstable system.  To bring the temperature
and  pressure back into thermodynamic equilibrium, the
temperature must be reduced to the pointwhere it is in equilibrium
with  the  reduced pressure  of the system.  Heat is lost by
evaporation of the residual water and contaminants, which are
then removed  from the system by  vacuum extraction.  The
injection of steam after equilibrium  has been reached at the
lower pressure allows a recharge of the heat necessary to drive
the evaporation of contaminants, andthedepressurization cycle
can then  be repeated.  Itamura and Udell (1995) show that
cycling of steam injection with continuous vacuum extration will
always reduce the amount of steam  required to meet a given
clean-up level, and may also  reduce the overall clean-up time.
  Another mechanism that can enhance contaminant recovery
during steam injection is enhanced desorption of contaminants
from the porous solids. For a contaminant to desorb from a solid
surface, it must absorb heat. The amount of heat that is required
for desorption to  occur is dependent on both the contaminant
and  the type of  soil.  Lighty et al.  (1988) have found that
essentially all of a semivolatile contaminant could be desorbed
from glass beads and sands, but only about 80 percent  of the
contaminant was desorbed at the same temperature from porous
clays and peat. The experimental results lead to the conclusion
that a monolayer of adsorbate is strongly bound to the solid
surface of a reactive soil such as a clay. The desorption of this
monolayer is a long process  (Tognotti et al., 1991), and likely
requires temperatures significantly above the boiling point of the
contaminant for complete removal (Lighty etal., 1988). Lighty et
al. (1990) found that xylene, even at temperatures above its
boiling point, can still adsorb from a gas stream onto a clay
particle. The very slow desorption indicates that a very strong
bond can  be  formed between reactive  soils and  organic
contaminants.
  Thus, contaminants can be recovered in the vapor phase, as
a separate phase liquid, and  dissolved in the aqueous phase.
The  relative amount of the  contaminants  in each  phase will
depend on the original concentration of the contaminant and its
boiling point.  Field demonstration projects  of steam injection
where gasoline and diesel fuel were recovered have found that
most of the contaminant recovered was recovered in the  vapor

-------
phase  (Newmark and  Ames,  1995;  EPA,  1995b),  while  a
demonstration to recover JP-5 recovered most of the contaminant
as a liquid (Udell et a!.,  1994).
  Although it has been shown that steam injection can potentially
recover a large percentage of volatile contaminants, it is expected
that  residual amounts  of the contaminants will remain  in the
subsurface. These small amounts of contaminants can likely be
remediated by natural attenuation or bioremediation. Thus, it is
importantto understand the effects of steam injection on microbial
populations and their ability to degrade residual contaminants.
Biological samples were taken at the Naval Air Station Lemoore
after a  steam injection demonstration to  recover JP-5. These
samples  showed high numbers of active bacteria in the zones
that  reached 80°C to 100°C for extended periods  during the
demonstration and suggested enhanced biodegradation was
occurring in the heated, oxygenated soil (Udell et  al., 1994).
Research carried out  as part  of the Dynamic Underground
Stripping project showed  that before  steam injection a wide
variety of microorganisms were actively degrading  the  BTEX
components of gasoline.   Pseudomonas was the dominant
species originally, but Flavobacterwas dominant after vacuum
extraction. Above the water table, the largest populations were
in areas  where the contamination was at low concentrations.
Below  the water table,  oxygen concentrations were  low and
there was effectively no  microbial activity. After steam injection,
extensive microbial  communities were found in  all samples,
including those where the temperature had reached 90°C. The
population, however, had shifted to yeasts and related organisms
which had been observed in small numbers before heating. The
community includes thermophiles previously  identified  from
environments such as hot springs, and a number of organisms
apparently  represent previously unidentified  species.   The
community includes the  BTEX degrader Rhodotorula (Newmark
and Aines, 1995).
Contaminant and Soil Type Considerations
  The decision to use steam injection for remediation should be
based on considerations of both the contaminant to be removed
and the properties of the porous media to be remediated. The
most important property of the  contaminant in  determining its
receptiveness  to remediation by steam is  its  volatility.   If  a
contaminant is not volatile, a hot water displacement without the
expense  of steam injection  can reduce  the saturation of the
contaminant to its residual saturation. For  contaminants that
have very low boiling points and a large Henry's constant, and
which are above the watertable, vacuum extraction may be able
to adequately remove a significant portion of the contaminant
from sandy, homogeneous soils. Steam injection is justified for
removing trapped lenses or ganglia of a volatile or semivolatile
contaminant that cannot be removed by  viscous forces.   Udell
and  Stewart (1990) feel that steam injection has perhaps the
greatest potential to significantly decrease clean-up time and,
therefore, offset the greater capital costs of the system when
semivolatile contaminants are  to be  removed.  Semivolatile
contaminants include the less volatile petroleum hydrocarbons,
such as diesel or jet fuels, and some of the higher chlorinated
solvents.  However, temperatures  significantly greater than
ambient temperatures may be required to desorb even volatile
contaminants,especiallyfromclaysorpeat Highertemperatures
enhance  the volatilization of liquid or  adsorbed contaminants
and  allow greater recovery (Lingineni  and Dhir, 1992).  Thus,
steam injection, along with vacuum extraction, may be necessary
to achieve the desired clean-up levels for volatile contaminants
when they are present  in fine soils.  Most of the volatile and
semivolatile organics show very significant increases in volatility
as the temperature increases, and most have very high vapor
pressures at steam temperatures.
  Stewart and Udell (1988) have shown theoretically that the
viscosity of the contaminant is important in determining the size
and saturation of the contaminant bank displaced by the steam
front Their calculations show that steam is capable of mobilizing
any length of contaminant bank up to a contaminant/water
viscosity ratio of three. For largerviscosity ratios, the displacement
is unstable; i.e., fingering may occur, reducing the efficiency of
the displacement.  Their  laboratory experiments support this
theory, showing that low viscosity contaminants can be recovered
in front of or just behind the steam front, while a mineral  oil with
a viscosity seven times that of liquid water at steam temperature
was not recovered by steam injection.
  Soil type has been found to have a strong effect on the rate of
contaminant removal from soils (de Percin, 1991;  Lighty et al.,
1988).  Adsorption onto glass beads and silica sands does not
appear to form tight bonds; the adsorption is  readily reversible
even at  low temperatures. Experiments performed by Lighty et
al. (1988) showed that essentially all ofthexylene adsorbed by
silica  sands was  recovered rapidly.   However,  for reactive
media, the desorption  process is  much slower, which may be
caused by a strongly adsorbed monolayer on the particle surface
(Lighty et al., 1988; Tognotti et al., 1991) or slow diffusion from
small inner pores to the surface of the particle (Keyes and Silcox,
1994).
  Laboratory studies  by Hadim et al. (1993) have shown that
remediation  by steam injection can be achieved much more
rapidly in coarsersoils than in fine soils. Their experiments have
also shown that soils which are poorly sorted (i.e., have a wider
range of grain sizes) have a lowercontaminant recovery efficiency
when  steam is injected.  This effect  is due in part to lower
permeability in fine grained or poorly sorted soils, which causes
a lower injection rate at a given injection pressure, and a greater
residual saturation. For sands within a certain range of grain
sizes, the same contaminant recoveries were achieved, but as
the grain size decreased, longer treatment times were required
to achieve the same recovery. As the grain size was decreased
further, there was a reduction in the recovery efficiency.
  Heterogeneity in  the subsurface can have  a very significant
effect on  the efficiency  of steam injection.  Small scale
heterogeneities can cause fingering, especially at the higher
flow rates (Basel and Udell, 1989).  Channeling will occur when
layered  heterogeneity exists due to naturally occurring beds or
to  manmade  disturbances of the  subsurface such as  highly
permeable trenches containing buried utilities (EPA,  1995b).
The degree of channeling  that will occur increases  as  the
differences in the permeabilities of the  layers  increases, as the
ratio of the layer  thicknesses  (more permeable  layer/less
permeable layer) increases, and as the rate of steam injection
increases (Basel and Udell, 1991). The occurrence of channeling
can allow contaminants in the low permeability zones to be
bypassed by the steam. However, there have been times when
a relatively impermeable layer overlaid the target zone for steam
injection, andthe impermeable "cap" limitedthevertical movement
of  the steam  (i.e., steam override),  and thus increased its
horizontal spreading (Farouq All and Meldau, 1979; Aines etal.,
1992).   Thus, although lenses of less permeable soils that
contain contaminants can reduce the efficiency of steam injection,
full layers of this low-permeability medium overlying the permeable
zone to be treated may actually improve the sweep efficiency of
the injected steam through the permeable zones.

-------
  Many researchers have found that the recovery from steam
injection is related closelyto heating rates (Myhill and Stegemeier,
1978; Baker, 1969; EPA, 1995b). However, this dependence is
more related to steam properties and injection rates; the thermal
properties of soils and the  liquid contaminants have generally
not been found to vary enough to affect the recovery achieved by
steam  injection (Myhill  and Stegemeier,  1978; Hadim et al.,
1993).  One property of the soil or aquifer to be treated that may
significantly affect the process is the thickness of the target/one.
In any steam injection process, some of the heat that is injected
will be lost to the overlying and underlying strata. Thethickerthe
target zone to be treated,  the greater the  percentage of the
injected heat that stays  within this zone.
  Another property of the media that can affect the efficiency of
the heating process is the permeability. When the permeability
is  low, the injection  rate at a given  pressure is low, which
increases the time for heat to be lost to  the overburden and
underburden, and decreases the efficiency of the overall process.
Thus, sandy media are  more easily treated by steam injection
than clays.  Fan and Udell (1995) studied the movement of the
evaporation front during steam  injection from areas  of high
permeability into beds of lower permeability. They found that the
iowerthe permeability of the less permeable zone, the higherthe
temperature must be in that zone for evaporation to occur. The
reason for this is the decrease in vapor phase flux in the  zone
with  decreasing permeability, creating a higher pressure in the
zone and thus slowing evaporation.
  Physical factors  relating  to the location  and quantity of the
contaminants in  the  subsurface are  also  important.   Small
volumes of contaminated  soils  may be more  economically
treated by excavation and  incineration, especially if the
contaminated soil is at or nearthe ground surface.  The pressure
of the  injected steam is limited  by  the overburden pressure,
which is a function of the depth of soil above the zone of injection.
When the steam pressure is greaterthan the overburden pressure,
fracturing of the overburden may occur, which could allow short-
circuiting of the steam  to the  ground  surface. Thus, shallow
contamination may severally limit the use of steam injection as
it limits the injection pressure that can be used, again pointing to
the use of excavation and incineration  or thermal desorption for
contaminants nearthe soil surface. The steam injection method
may be a very efficient way to treat contaminated soil at significant
depths, and can be used to depths greaterthan 100 feet When
the contamination is contained in a very narrow depth range, and
there is no overlying and underlying  confining layers to limit
steam  movement  in the vertical direction,  steam injection
efficiency may decrease (EPA, 1995b). Steam injected into a
fractured media would  be expected to flow through the  more
permeable  fractures and would be ineffective for displacing
contaminants trapped in the porous matrix or in deadend fractures.
However, conductive heating from the fractures into the matrix
may effectively distill trapped volatile contaminants which could
then flow in the vapor phase to recovery wells (Udell, 1997).

  When  liquid phase contaminants are  displaced, they are
concentrated in front of the steamzone as the steam displacement
progresses.  This concentrating  of liquid  phase contaminants
may create a potential for  downward migration  of the liquid
phase contaminant, particularly if the liquid phase contaminant
is more dense than water. A confining layer beneath the zone to
betreated may help in limiting liquid phase contaminant downward
movement during steam injection (EPA, 1995b).
Design Considerations
  One of the most important design considerations for a steam
injection process is the steam injection rate. Thisfactor, however,
cannot be separated from considerations of the injection pressure,
temperature, and steam quality. The injection pressure is limited
by the depth of injection: when injection pressures greaterthan
the overburden  pressure are used, fracturing can occur in the
overburden, allowing steam to escape to the surface. Udell etal.
(1994) recommend that the injection  pressure be as high as
possible while not exceeding the soil fracture pressure which is
estimated as 1.65  psi per meter of depth  below the  ground
surface. Thus, the shallowerthe zone to be treated, the Iowerthe
injection  pressure  must be.  There  is a  direct  relationship
between  the injection  pressure  and injection rate, which are
related by the permeability of the media. The more permeable
the media, the greater the injection rate that can be achieved at
a given limiting  pressure.  The greater the injection rate, the
greater the heating rate of the subsurface. Newmarkand Aines
(1995) recommend large amounts of steam for establishing a
complete steam zone in very permeable media.  However, once
the steam has reached the production well, the higherthe steam
injection rate the  more steam (and therefore heat) that is produced.
Thus, once steam breakthrough  has occurred, greater thermal
efficiency can be achieved by reducing the steam injection rate
(Myhill and Stegemeier, 1978).   Newmark and Aines (1995)
found that repeated steam passes were effective for heating
small impermeable layers in between steamzones. The optimum
injection rate is dependent on many variables specific to a given
system, such as distance between wells, sweep efficiency, and
heat losses to  over- and  underburden, and may  be best
determined by field experimentation (Bursell et al., 1966).
  Greater heating rates generally mean greater recoveries, and
greater energy  efficiency.  Increasing the temperature of the
injected steam, however, does not necessarily increase the rate
of heating of the target area.  Johnson et al. (1971), studying
steam displacement of oil from a reservoir, found that when they
increased the temperature of the steam they ultimately recovered
more oil, but the greater temperature gradient  between the
steam injection zone and the over- and underburden increased
the rate of heat loss, thus the steam front actually advanced more
slowly.  Longer  times were required to achieve the increased
recovery,  and  the  additional oil recovered  at the higher
temperatures required a  disproportionately larger amount of
steam. Baker (1969) found that for a given steam injection rate,
the rate of heat losstothe overburden and underburden increases
with time as the area of contact with them grows. At some point,
the rate of steam input will equal the rate of loss to the over and
underburden, and the growth of the steam zone will stop. Thus,
there is a maximum size to the steam zone for a given injection
rate.
  Several researchers have found a direct relationship between
the steam quality and the oil/steam ratio at the producing well.
Steam quality is defined as the proportion of the total water that
is in the vapor phase;  if liquid water is not present, the steam
quality is  100 percent.   Myhill  and  Stegemeier  (1978)  and
Singhal (1980)  found that as the quality of the steam at the
injection point increases, the oil/steam ratio in the extraction well
increases.
  Pilot- and full-scale  steam injection demonstration projects
seem to indicate that  greater efficiency is achieved by using
continuous steam injection; i.e., 24 hours a  day and  at least 6

-------
days per week, rather than intermittent injection (16 hours per
day, 5 days perweek). The intermittent injection schedule allows
the soil to cool somewhat during the 8 hours when steam is not
being injected, significantly slowing the overall heating rate and
maximumtemperaturesinthetargetzone. Intermittent operation
also was found to put additional stress on the boiler and other
process equipment due to the frequent cooling and heating, and
caused additional system down time (EPA, 1995b).
  A physical  characteristic of steam injection  processes that
must be considered is steam override due to gravitational forces.
Gravity override is caused by the  fact that steam is much less
dense than liquids and therefore will tend to rise in the porous
medium.  At  the same  time, hot  water  produced  by  the
condensation of steam is more dense than some contaminants,
so maytendto underridethezoneto be treated (Singhal, 1980).
A frontal displacement  mechanism with essentially pistonlike
flow will displace more of the initial  fluids by viscous forces.
When steam  override occurs, the effectiveness of the frontal
displacement is reduced, and thus the vertical sweep efficiency
of the steam displacement process is reduced. Various properties
of the system determine the amount of override that occurs.
When steam is injected into an unsaturated porous  medium the
steam front is essentially vertical.  When there is a difference in
density between the  injected fluid  and the fluid initially present,
the slope  of the interface between the fluids  is equal to the
difference  between the  vapor and liquid water phase viscous
forces divided by the gravitational forces. The degree of gravity
override increases as the difference in density between the liquid
and vapor  phases increases, as the permeability of the medium
decreases, and  as  the viscosity of the liquid phase being
displaced  increases. One operational parameter  that affects
override is the steam injection rate.  Steam override cannot be
eliminated, but by increasing the injection rate of the steam, the
difference betweenthevaporand liquid viscousforces is reduced
and thus the amount of override can  be decreased (Basel and
Udell, 1989).
  The  major equipment requirements for  a steam  injection
system are the steam generator, the  distribution system to the
wells, the extraction system, and the coolers/condensers forthe
extracted fluids. Means of treating the off gases, the water that
is extracted, and any organic phase that is  recovered will be
required, but these systems are beyond the scope of this paper
and will not be discussed further.   Mobile steam plants are
available (Newmark, 1992) that are  powered by natural gas,
propane, or another available fuel.  Steam generators require a
high quality feed water to  avoid scale buildup in the generator
(Schumacher, 1980), so often the generator feed water must be
treated before use.  Normally, operation of a steam  injection
system will require that the boiler be operated and  manned
continuously during the  injection process (Newmark,  1992).
  Pneumatic  air lift  pumps (Newmark, 1992) and jackpumps
(Udell and  Stewart, 1989) have been used in the extraction wells
to remove  liquid water and contaminants.  Blowers are used to
remove steam and  contaminants  in the vapor phase.  Heat
exchangers are used to condense the extracted vapors and to
preheat the steam generator feed water.
  Special consideration must be given to the construction of the
steam injection wells  due to the high temperatures and pressures
they must withstand.  PVC or fiberglass wells that are commonly
used for ground-water monitoring are not adequate for steam
injection processes.  Steel casing is commonly used for both
injection and extraction wells, but the installation must allow for
the expected expansion of  the  casing  with temperature.
Experience in the oil industry has shown that pipe failures in
steam injection projects occur at the couplings, due to repeated
expansion  and contraction during the heating  and cooling
processes  (Gates and  Holmes,  1967).  Cement  used for
conventional wells will frequently fail when exposed to the high
temperatures associated with the steam  injection  process
(Schumacher, 1980). The addition of 30 to 60 percent by weight
of quartz silica or silica flourto conventional cements will provide
temperature stability, and the addition  of sodium chloride will
allowthe cementto expand linearly with temperature (Gates and
Holmes, 1967). Information on well construction casing materials
and  cements that have worked for steam injection wells is
available from both the oil industry (Gates and Holmes, 1967;
Hall  and Bowman, 1973; Farouq Ali and Meldau, 1979; Chu,
1985) and  from pilot-scale  contaminant recovery operations
(Newmark,  1992).
  The  placement of steam injection  and  extraction wells is
critical to the efficiency of the remediation system.  Determining
well  placement and  injection zones should be  based on a
thorough knowledge of the subsurface characteristics, including
delineation  of zones  of high permeability and high contaminant
concentration. If the target area to be remediated is nottoo large,
injection wells  may circle the contaminated zone with one or
more extraction wells in the center.   When  the area to be
remediated is  large  enough that  the  desired  injection rates
cannot be achieved  by peripheral injection wells,  injection and
production wells are usually arranged in a pattern. The 5-spot
pattern (see Figure 3) alternates injection and extraction wells so
that all wells are surrounded by four wells of the opposite type.
Patterns using two injectors per  producer (7-spot pattern, see
Figure 3) have  also been used  by the oil industry to recover
viscous  oils (Boberg, 1988).  Patterns  such as these may be
effective when the subsurface and contaminant distribution is
essentially homogeneous. When significant heterogeneity exits,
well placement musttake this into account. When a pronounced
directional  permeability trend exists, this will  control well
placement.  Edge or  peripheral drives may be more appropriate
when there is significant geologic structure (Willhite, 1986).
Farouq Ali and Meldau (1979) recommend close well spacings
and an ample number of production wells, and Udell etal. (1994)
found at a field  demonstration that 5 to 6 meter well spacings
appeared to work better than 12 meter well spacings.  Based on
A   o  A   o
  \     /  \   x

 oV o  *
   .^s     f  "•>
 <  o  V   o  XAX
  \   /  \     /  \
    \u     \  /
 o  A   o  X  o
     Five
  A--A   o    A--I

's   O   A — ^   o
 \     /       \
           O   k---^
•x   o    o,~i^   o
 \     /      \
  cr - -A.   o   it- ~ -
                      A  Injection Well
                      o  Production Well
                  - - - -  Pattern Boundary


Figure 3. Fivespotandsevenspotwellpatternsusedforsteamflooding.

-------
the results of this demonstration, they also recommended that
flexibility be built into the system by installing wells that can be
used for either steam injection or fluid extraction.  The heat
losses observed by Baker (1969) would suggest that there is a
maximum  radial area that can be heated by a given steam
injection rate,  and this  would set an upper limit on distance
between wells  in order to heat the entire area.  Pilot studies for
contaminant recovery have used well spacings as  low as 1.5
meters, but full scale operations have used well spacings on the
order of 18 meters (Aines et  al., 1992;  Newmark,  1992; EPA,
1995b).
  The sweep efficiency  of the injection process is the areal and
vertical amount of the formation targeted for treatment that is
actually  contacted by the steam.  Sweep efficiency can be
reduced by areas of lower permeability that are bypassed by the
steam.  When large areas  of the formation contain  lower
permeability materials, steam may essentially miss the whole
area. Onetechniquethathas been used successfullyto increase
sweep efficiency is to shut down the production wells which have
a good connection to the injection well once those areas are
clean (i.e., are no longer producing contaminants), and to
continue producing from wells that are in the direction in which
steam penetration is desired (Powers etal.,  1985; EPA, 1995b).
Where a high viscosity fluid is to be recovered it may be desirable
to heat a large  portion of it early on to reduce its viscosity.  This
has been accomplished by injecting steam into a  lower more
permeable layer, and allowing heat to transfer upward to the oil
and increasing its mobility (Hall and Bowman, 1973). In a case
where a viscous oil was floating on top of the watertable, steam
injection belowthe watertable was found to conduct heat into the
oil layer to increase its mobility (Farrington and Sword, 1994).
Gravity override may also aid  in distributing heat to fluids of low
mobility. Steam may spread evenly on top of a contaminant layer
and conduct heat downward to heatthe contaminant and increase
its mobility through viscosity reduction and/or distillation (Farouq
Aliand Meldau, 1979).
  Effluent stream monitoring is required to monitorthe progress
of the remediation process. Vapor and aqueous phase samples
are normally collected at regular intervals during the course of a
remediation and analyzed to determine the  amount of
contaminants being  removed.  Sample  intervals of one  hour
have been used on a small-scale demonstration project, while a
one-day sampling interval was used on afull-scaledemonstration.
However,  the  time  required to analyze  these  samples by
techniques such as gas chromatography limits their usefulness
for process control and optimization purposes. Also, it has been
found that effluent streams, particularly the vapor effluent stream,
can vary significantly over short periods  of time,  and these
variations cannot be monitored  with the one-day or even one-
hour periods between samples that are generally used for grab
samples. Flame ionization detectors have been used with some
steam  injection systems  for  real time  monitoring  of the
contaminants being recovered in the  vapor phase (de Percin,
1991; EPA, 1995b). Fourier transform infrared  radiation (FTIR)
(Langry  and Kulp, 1994) and differential ultraviolet  absorption
spectroscopy (DUVAS) (Barber et al., 1994) have  both been
evaluated  for real time monitoring of effluent  vapors.  FTIR
measures alkane components in the vapor stream, while DUVAS
measures  aromatic compounds.  To date, the  concentration
information fromthese on-line monitors has been more qualitative
than quantitative,  but these monitors  have been shown to
provide  a real time estimation of the  amount of hydrocarbons
being recovered by the system. In orderto correctly interpretthe
effluent concentration data, it must be kept in mind that a lag time
between injection and the effect on the extraction well is to be
expected.  Also, mass transfer limitations can occur within the
subsurface, which reduces effluent concentrations and makes it
appear that the concentration remaining in the subsurface is at
lower levels than it actually is. Thus when the extraction system
is shut down for  a  period of time and then restarted, effluent
concentrations are higher than they were before the shutdown.
  Tracking the movement of the steam  injection front in the
subsurface is also desirable for monitoring the progress of the
process, and to aid in understanding the processes that are
occurring  as a  result  of  steam injection.  Temperature
measurements, eitherat intermediate observation wells oratthe
producing  wells, provide a direct means of tracking steam front
movement (Hall  and Bowman,  1973; Powers  et al., 1985;
Newmark,  1992). However, when additional steam passes are
used after the subsurface is heated to approximately steam
temperature, the small differences in temperature produced by
the passage of an additional steam front may be hard to detect.
Thus, other means of tracking steam front movement may be
needed (Newmark and Aines, 1995).  Powers et al. (1985) also
monitored  the chloride concentration  in the naturally  occurring
brine that was produced, and found measurable decreases in its
chloride content due to dilution by condensed steam before  a
thermal response was detected.  Neutron logs will indicate the
presence of vapor saturation and, thus can be used to monitor
the steam front movement (Hall and Bowman, 1973). Radioactive
tracers have also been used to determine which injection wells
are influencing each of the various producers and to what extent
(Powers etal., 1985).
  The DynamicUnderground Stripping Project(Newmark, 1992;
Newmark and Aines, 1995) used a wide variety of geophysical
techniques both  before steam injection to establish baseline
information and to help characterize the subsurface, and during
steam injection to test their ability to monitorthe movement of the
steamfront. Geophysicallogswerecomparedtothetemperature
records  obtained from monitoring wells to provide additional
insight into the  results  from the  geophysical logs.   Low
permeability, clay-rich zones have relatively large amounts of
exchangeable cations, giving them a relatively lower electrical
resistivity than more coarse-grained soils. Electrical resistance
tomography (ERT) provides  cross-sections of subsurface
resistivity and can identify the higher permeability zones where
steam flow is likely to  occur.   Temperature increases  may
increase the mobility of exchangeable cations and thus decrease
resistivity,  or in areas where the steam has caused desaturation,
the  resistivity may increase (Vaughan  et al.,  1993).  Thus,
changes in resistivity were found to be useful for mapping the
movement of the steam front (Ramirez et al., 1993).  Induction
resistivity logs measure the resistivity of the pore fluids nearthe
boreholes  providing a more detailed vertical view of steam in the
subsurface, and were found to delineate the vertical temperature
distribution.  Tiltmeters,  which  measure deformations in the
ground surface that result from a  pressure transient in terms of
tilt, also showed potential for determining the movement of the
steam front. Tiltmeters were found to be more useful after the
ground was already heated to approximately steam temperature
to map the movement of additional steam passes (Newmark and
Aines, 1995).
  Numerical models may be useful as an aid to designing clean-
up operations using steam injection, and to provide estimates of
the  time required to achieve different levels of contaminant
removal (Newmark, 1992).  Many models of heat flow (Spillette,
1965; Marx and Langenheim, 1969; Prats, 1969) and oil recovery
from a reservoir due to  steam injection (Willman et  al., 1961;

-------
Vinsome,  1974; van  Lookeren,  1983; Rubin  and Buchanan,
1985) have  been  developed.  Because there are  significant
differences between  steam injection for oil recovery and for
contaminant removal, numerical  models developed  for oil
recovery processes would not necessarily include all of the
processes that are important in  the contaminant remediation
process.   Partitioning processes between  the air, water, and
solid phases that occur at the  pore level are  not generally
important in oil recovery operations, but may be very important
when considering removing contaminants down to the parts per
million  and parts per billion level (Falta et al.,  1992a).  The
thermal model developed by Rubin and Buchanan (1985) is
general in its formulation and may be adequate to model some
aspects of a steam injection process for subsurface remediation.
Afinite-differencesimulatorcalled the Multicomponent Multiphase
Nonisothermal  Organics Transport Simulator (M2NOTS) has
been developed by the University of California at Berkeley to
simulatethe steam injection process forthe purpose of subsurface
remediation. This simulator was used to aid in the design of the
Dynamic Underground Stripping Project, however, comparisons
between simulations and the field trial at the gasoline spill site
are not currently available.

Laboratory Experiments

  Two-dimensional laboratory experiments on steam injection
have also been done. When  steam was  injected into a two-
dimensional homogeneous sand  pack containing o-xylene, the
xylene was displaced as a free product bank ahead of the steam
condensation front.  Essentially  complete  recovery  of the
contaminant was achieved with the injection of approximately
350 pore volumes of steam. However, when steam was injected
into a layered sand pack containing diesel fuel, there was not a
significant physical displacement of the diesel by the steam
condensation front Instead, the volume of the initial mobile pool
decreased as it was displaced  by the steam front, and residual
hydrocarbon liquid was observed in all regions that had contained
the diesel.  Continued steam injection and vacuum extraction
then produces a fractional distillation of the diesel fuel components
with the most volatile components being removed first (Udell,
1994).
  Itamura and  Udell (1993)  studied  the  recovery of
tetrachloroethylene (PCE),  a dense, semivolatile contaminant,
from a two-dimensional sand packthat contained mainly layered
heterogeneity.  Near the top of the model and above the water
table was a layer containing diamond-shaped lenses of varying
permeability sands. The contaminant was introduced at the top
of the  model and the contaminant that reached the zone of
variable permeability lenses preferentially migrated to the lowest
permeability zones in this region and then along the boundaries
between the different permeability sands.  Steam was injected
over most of the vertical depth of the model, invading the most
permeable layer first, and  then the variable permeable layer
containing most of the contaminant. When the injected steam
reached the contaminant, it was displaced both downward and
horizontally towards the recovery well. The  mobile contaminant
in  front of the condensation front moved  along  the boundary
between different permeability sands. Afterthe mobile, separate-
phase  contaminant  had  been  displaced, the contaminant
continued to be recovered in the vapor phase.  The highest
recovery rate of PCE occurs just  before steam breakthrough in
the regions containing the majority of the contaminant. Visually
it could be determined that some of the contaminant remained in
the model, butthe percent of the contaminant recovered was not
reported.
Field Trials
  Table 1 contains a compilation of some of the details of steam
injection field trials that have been carried out. In The Netherlands,
steam stripping has been used to remediate contaminated soils
since at least 1983.  However, only limited information on  the
applications of steam stripping could be found  in the literature.
Hilberts (1985) describes a vacuum bell structure that is placed
on top of the soil and is commonly used forthe steam injection
process. The 2 meter by 2 meter box contains 4 lances for steam
injection, and has been used to treat contaminated soils as deep
as 4.5 meters. Contaminants are drawn off from the center of the
bell, which creates a vertical flow of the steam up through the soil.
At the Broomchenie site which  contained organic bromides,
reductions of 97 percent  in the most heavily contaminated  soil
were found, but further analysis showed that most of the organic
bromide had been converted to inorganic bromide which remained
in the soil. Only a small amount of bromide was present in the
condensate removed from the soil (Hilberts, 1985).
  A small-scale pilot demonstration project was  undertaken at
the Solvent Services, Inc. site (San Jose, California) in August
1988  (Udell and Stewart, 1989). Surface  spills and leaking
underground storage tanks at this industrial facility had released
a variety of volatile organic compounds to the subsurface.
Extensive temperature monitoring showed that the movement of
the steam zone  was controlled.  Cycling of vacuum extraction
and steam  injection was then  carried out sporadically after
steam breakthrough and significantly greater recovery rates
were achieved at the beginning of the second vacuum extraction
period.  Comparison of soil core analysis done  before and after
the pilot remediation  study  indicates that there  was  some
downward  migration  of contaminants  in  solution  in  the
condensate.  Low permeability zones isolated contaminants,
and the post-treatment concentrations in some of these zones
was higherthan the initial concentrations. The authors postulated
that high water  saturations containing high concentrations of
contaminants which were pushed in front of the steam front were
imbibed into the low permeability areas and caused the observed
increases in the more highly water-soluble contaminants. Overall,
the pilot study demonstrated the potential for steam injection in
conjunction with vapor  extraction as a  rapid  and effective
remediation technique.

  de Percin (1991)  reported  on a demonstration  project for the
"Detoxifier"  system  made by NovaTerra, Inc.  This system is
similarto that used in The Netherlands in that it is a shroud or bell
placed on the soil surface in which a vacuum is  maintained to
remove the contaminant vapors.  In this case, instead of lances
to inject steam, the steam, along with hot air, is injected through
augers that are  rotating in opposite directions  to break up  the
soil.  The steam and hot gases carry the  contaminants to  the
surface. The augers can treat 2.5 m3 of soil to a depth of 8.2
meters. Blocks  of soil  are treated sequentially, and treatment
time  is  varied dependent on the soil type and contaminant
concentration. Aflame ionization detector measures the organic
concentration of the gases  in the shroud, which are roughly
correlated to the concentration of organics remaining in the soil,
and can indicate when a desired level of treatment has been
achieved. A tracer study showed that the mixing action of the
Detoxifier auger does not produce a homogeneous treatment
area, thus all of the soil may not receive adequate treatment.

  Basile and Smith (1994) reported on a combined anaerobic
degradation/steam  injection/vacuum extraction  system  being
used to remediate  separate  and aqueous phase  chlorinated
hydrocarbons. Laboratory data showed that elevated bioactivity

-------
Table 1.  Summary of Steam Injection Projects for Subsurface Remediation.
Site/
Reference
Utrecht's
Griftpark,
Netherlands
(Hiiberts, 1985)



Broomchenie,
Wierdin,
Netherlands
(Hiiberts, 1985)
Mannheim,
Netherlands
former gas
works
(Hiiberts, 1985)
Solvent
Services, Inc.
(Udell and
Stewart, 1989)




Contaminant
Concentration/
Volume
BTEX, naphthalene,
PAH, phenol





Organic bromide
compounds, 3 -7700
mg/kg

Benzene 55 mg/kg
Toluene 15 mg/kg
Xylene & Ethylbenzene
2-4 mg/kg
phenol 30 mg/kg
VOCs and nonvolatile
organic contaminants
at concentrations
greater than 1000 ppm




Description of
Geology
layered sands,
slags, clay and
bog




Sand, ground-
water at 5 m
depth

Rough sand
material



Silts and clays,
continuous
poorly sorted
sand layer at
bottom, 0.61 to
1.5m


Treatment System
Design







Treatment to depth
of 4.5 m


Treatment at depth
of 1.8 to 2.6 m



7.3 m2 area treated
by 6 Injection wells,
1 extraction well, 1.5
m between wells,
111.6 kg/hr of steam
injected for 120 hrs,
then 67 kg/hr for 20
hrs
RemoYal Efficiency Comments
or Volume
BTEX: 99.5% (sand), Vacuum bell
20% (clay);
naphthalene: 99.9%
(sand), 60% (clay);
PAH: 97% (sand), 35%
(bog); phenol: 80%
(sand), 20% (day)
97% Vacuum bell
Converted to inorganic
bromide

All contaminants Vacuum bell and steam
reduced to below drive
detection limits


Vacuum Extraction: 99 Pilot scale
kg in 40 hrs demonstration
Steam extraction: 146
kg in 140 hrs
Followed by
intermittent operations


                                                                                                                 Continued

-------
Table 1.  (continued)
Site/
Reference
Solvent
Services, Inc.
(Udell and
Stewart, 1989)




Annex Terminal
Port of Los
Angeles, San
Pedro, CA
(de Percin,
1991; EPA,
1991)
AT&T
New York
(Basile and
Smith, 1994)

Contaminant
Concentration/
Volume
VOCs and nonvolatile
organic contaminants
at concentrations
greater than 1000 ppm




Major contaminants
wereTCE, PCE,
chlorobenzene
Initial average
concentration of 466
ppm VOCs

Chlorinated solvents,
TCE&1.1.1-TCA,
DNAPL (separate and
dissolved aqueous
phase)
Description of
Geology
Silts and clays,
continuous
poorly sorted
sand layer at
bottom, 0.61 to
1.5m









Tight,
heterogeneous



Treatment
Design
7.3 m2 area treated
by 6 injection wells,
1 extraction well, 1.5
m between wells,
111.6 kg/hr of steam
injected for 120 hrs,
then 67 kg/hr for 20
hrs
33 m3
to 1 .5 m depth by
steam (200 °C) and
compressed air
(135°C)







Removal Efficiency
or ¥olume
Vacuum Extraction: 99
kg in 40 hrs
Steam extraction: 146
kg in 140 hrs
Followed by
intermittent operations


84.7% VOCs, 55%
SVOCs
Treatment time was 1
month



4,500 kg of
hydrocarbons
recovered in 2 years


Comments
Pilot scale
demonstration






Detoxifier system
developed by
NovaTerra, Inc.




Nutrients injected at
40° C, vacuum
extraction of 7.8
m3/min, maximum
vacuum of 41 cm Hg
                                                                                                                                Continued

-------
Table 1.  (continued)
Site/
Reference
Yorktown Naval
Shipyards
(Farrington and
Sword, 1994)








Naval Air
Station
Lemoore, CA
(Udell et al.,
1994; EPA,
1995a)


Pinellas Plant
Northeast Site,
Largo, FL
(DOE, 1997)
Contaminant
Concentration/
Volume
Naval Special Fuel Oil,
estimated 8000 I1










JP-5, estimated
757,000 I






Volatile Organic
Compounds including
BTEX and chlorinated
solvents
Description of
Geology
Upper 6 m was
homogeneous
fine to coarse
sand, below that
was interbedded
sands and clays.
Water table 3.8
to 4.1 m below
ground.
Hydraulic
conductivity 2.0
to 5.2 x 1Q'3em/s
Sands and silts
with hydraulic
conductivity of
a.gxIO^to 1.4x
10"2 cm/s. Water
table at 4.9m


Silty sands,
water table at 1
m below ground
surface
Treatment
Design
83.6 m2 treated with
5 spot pattern of 4
injection wells, 1
extraction well, 9.1
m between injection
wells. Injected at
6.1 to 7.6 m depth,
extracted from 3 to
9.1 m depth.
Injection rate of 272
kg/hr

12, 140m2 treated
with 2 injection wells
at the center, 8
vapor/ground-water
extraction wells.
injection depth of
6m

1566 m3 treated by
48 holes to a depth
of 9.8 m

Remowal Efficiency
or Volume
6171 recovered
Steam injected over 2
month period









Approximately 976,000
I recovered in 3 months
of operation. Final
vadose zone
concentration of 20 to
SOppmTPH, 20,000
ppm remains at water
table
Approximately 544 kg
recovered


Comments
Hot water may be more
appropriate for this
nonvolatile, viscous oil









Demonstration project







Dual Auger Rotary
Steam Stripping
Demonstration project

  1 -The pilot study report does not contain an estimate of the amount of oil contained in the are a treated by the pilot study.  This estimate was
  made using the same assumptions as made in the report to estimate the oil contained within the entire contaminated region, with an estimate of
  the contamination extending over a 1.5 m depth.

-------
occurred at temperatures between 30°C and 40°C, so nutrients
are injected at 40°C to promote biological dechlorination of the
contaminants. The authors claim that the degradation products
are more easily extracted than the chlorinated solvents, and are
then steam stripped  and  removed from the subsurface via
shallow vacuum  extraction wells.  They claim that the  in situ
remediation was  favored over excavation and off-site disposal
by local residents because it eliminated the need to transport
contaminated soil through the neighborhood.
  The Dynamic Underground Stripping Project is a joint venture
between the University of California-Berkeley and Lawrence
Livermore Laboratory. The process combines steam stripping in
permeable layers and electrical resistance heating of clay layers
to heatthe subsurface.  Characterization of the site before steam
injection indicated  that about half of the contaminants were
above,  and half were  below the  water table.   The highest
gasoline concentrations were  found in the capillary  fringe
(Newmark and Aines, 1995). Electrical resistance heating of the
clay layers was  initiated at night (to limit interference  with
construction activities going on during the day) for 12 weeks prior
to steam injection, raising the temperature in the clays from 20°C
to 50°C. Steam was injected both  above and below the water
table, and during 5 weeks of continuous steam injection over
6,400 liters of gasoline was recovered, most of it in the vapor
phase.
  After about 3 months, the second phase of steam injection was
begun.  During this  phase, steam was injected  intermittently, so
that the treatment zone was kept desaturated and periodically
depressurized. Cycle times were 5 to 6 days, and this phase was
continued for 6 weeks.  Extraction of gasoline was found to be
particularlyvigorous when the treatmentzonewas maintained in
a depressurized condition at elevated temperatures, and a total
of 18,500 liters of gasoline were recovered.  Soil cores taken
after the second phase of steam injection showed that most of
the gasoline remaining in the subsurface was trapped in the low
permeability zone between the two  steam zones.  A final phase
of vapor extraction with limited  electrical heating and periodic
fluid extraction was  conducted for 3 months, which recovered an
additional 3,800 liters of gasoline (Yowetal., 1995).  It is believed
that no additional free gasoline remains in the treatment zone,
but benzene  and toluene remained  in the ground-water at
concentrations  above their maximum  contaminant  limits.
Maximum contaminant limits  had  also  been  set for  1,2-
dichloroethane and xylenes,  and these  levels had been met.
Biological sampling at the conclusion of the project showed that
BTEX degraders had  survived  the heating and  could rapidly
remove the remaining contaminants from ground-water(Newmark
and Aines,  1995).
  As part of EPA's SITE demonstration program (EPA, 1995b),
a steam injection  demonstration was carried out at the Rainbow
Disposal site in Huntington Beach,  California,  between August
1991 and August 1993. The soils treated by this system were
under and around existing facilities at the transfer facility, and
transfer activities continued during the remediation efforts.  The
discontinuous clay layers  at this  site created  a nonuniform
distribution  of the contaminant in the subsurface.  The spilled
diesel flowed downward through the  sand layers,  and when it
reached a  sand/clay  interface,  it  flowed laterally along the
interface until it reached a break in the clay layer which allowed
itto flow downward by gravity again.  This caused large variations
in concentrations even over small vertical distances.
  Maintenance of the boilers, the oxidizing unit to treat offgases,
and the steam distribution system caused considerable downtime,
and operating logs from the project showed the system had an
online factor of 50 percent. Initially the steam injection system
was operated 16 hours per day, 5 days per week. This led to
inefficient heating of the soil and maintenance problems with the
boilers due  to the thermal stress  associated  with  frequent
startups and shutdowns. An increase in the soil heating efficiency
and diesel recovery was noted when a 24 hours per day, 6 days
per week schedule was adopted, and the operation of the boilers
improved.
  The  clean-up criteria set by the California Regional Water
Quality Control Board  was 1,000  mg/kg of total petroleum
hydrocarbons (TPH).  It appears that this criteria was not met, as
45 percent of the post-treatment samples inside the treatment
zone were above this level. For the  small number of soil cores
available for which there is pre- and  post-treatment data, it
appears that the diesel  fuel  moved downward  in the time
between coring and steam injection.  The post-treatment soils
data shows the majority of the contamination is in the range of 8
to 12.2 meters in depth. This could have been caused by the loss
of a perched water table due to the very dry conditions in the
region. Steamwas injected at 10.7to 12.2 meters below ground
surface, and the 7.6 meter long screens on the extraction wells
pulled vapor and liquids from the 3 to 10.7 meter depth range.
Thus, the steam had a vertical component of flow, caused by the
vertical positions of the  screens on the  injection and extraction
wells. The growth of the steam zone was monitored by a limited
number of temperature wells, and this data shows that the region
from 6.1 to 9.1 meters below  ground  surface was  generally
heated to steam temperature.  At the bottom of the wells (12.2
meters), the temperature generally did not rise above 70°C.
Some of the highest  concentrations of TPH were in an area
where  underground storage tanks existed, and  this area only
received steam for a short period of time at the end of the project.
Thus, it appears that a majority of the remaining  contamination
(but not all) was in areas that did not reach steam temperature.
Also, cycling of steam injection and vacuum extraction, which
was found to be very  effective at other field sites for recovering
contaminants, was not used at this site.
   A more thorough understanding of the subsurface conditions,
a better designed injection/extraction system to heatthe lowest
zones  where contaminants  occurred, more temperature
monitoring wells throughout the treatment area and monitoring
of the steam injection  rate  at individual injection wells to monitor
the growth of the steam zone, and more effective operation may
have improved the overall effectiveness of this system.
  A pilot steam injection study was carried out at the Yorktown
Naval Shipyards  in Yorktown, Virginia (Farrington and Sword,
1994), where Naval special fuel oil had leaked from underground
storage tanks.  This oil is slightly less dense than water, with a
specific gravity  in the range of 0.94 to 0.99  at  ambient
temperatures, and is essentially insoluble in waterand nonvolatile.
The viscosity of this  oil, measured  on two  different samples,
ranged from 137 to 50  centipoise at 38°C. Thus, the oil at this site
is floating on top of the shallow watertable and moves only slowly
at ambient temperatures.  Steam was injected below the water
table to heat the oil,  and  oil began to  be recovered  from the
system when the average temperature reached about 60°C.
The viscosity of the oil at this temperature is in the range of 20
to 40 centipoise, a  substantial reduction from the  ambient
temperature viscosity. For an oil which is nonvolatile, and where
the injected fluid is not being used to physically displace the free
phase liquid, hot water injection, if implemented  correctly, may
be capable of recovering as much of the free-phase oil as steam
injection without the  associated difficulties and  extra costs of
generating and injecting steam.
                                                          12

-------
  A pilot scale demonstration project was carried out at Naval Air
Station Lemoore in Lemoore, California, in 1994 to recover JP-
5(Udelletal., 1994). The system consisted of two injection wells
surrounded by 8 extraction wells within the contaminated zone.
Thus, much of the recovered fluids, including  an estimated 80
percent of the recovered JP-5, was from outside of the targeted
treatment area enclosed by the extraction  wells. Concentration
of TPH inside  the treatment area in  the vadose zone was
reduced from as high as 100,000 ppm to 20 to 50 ppm. At and
below the water table, however, TPH concentrations remained
around 20,000 ppm. Soil samples taken after 35 days of steam
injection showed high TPH concentrations at the interface of the
surface clay layer and the underlying silty sand which had
previously appeared to be uncontaminated. Continued steam
injection reduced these concentrations. Recommendations for
reducing  or eliminating the upward spreading  of contaminated
vapors included control of the ground-water elevation to avoid
mounding during steam injection, higher JP-5 extraction rates in
both the liquid and vapor phases, and greater energy input rates
to avoid  condensation.   These  recommendations  could  be
achieved by installing wells  made for both steam injection and
extraction of liquids and vapors, allowing greatersystemflexibility
and by continuous monitoring  of  ground-water elevations  to
optimize fluid removal rates.
  A demonstration of a Dual Auger Rotary Steam Stripping
system by In-Situ Fixation, Inc., (DOE, 1997) was carried out at
the Pinellas Plant Northeast Site  at Largo, Florida, starting  in
December 1996. The system used was similartothe previously
described "Detoxifier" system.  The remediation  goal of this
demonstration project was to  reduce high contaminant
concentrations  (500 to 5000 ppm) to  levels which  are  more
amenable to anaerobic bioremediation, which was the chosen
remedial  technique for this site.  Although 75 to 95  percent
removals of contaminants from soil and ground  water  were
generally achieved, the treatment goal  of 100  to  200 ppm
remaining was not generally met. In the more highly contaminated
areas, the system was severely limited by the vapor treatment
system which  could  not treat the large  quantities of volatile
organics that were released.  Thus, it is difficult to fully assess the
effectiveness of the auger/steam injection system for releasing
contaminants from soil and ground water because of the limitations
placed on systemoperationfromthe undersized vapordestruction
system.

Cost Information
  Cost information on innovative remediation methods is sparse
because of the limited experience with applying these methods
in the  field.  Because each site is unique  in terms of the
subsurface geology, the types and quantities of contaminants,
and in terms of other site characteristics that may affect an in situ
treatment technology such as steam injection, costs incurred
during  the  application of this technology at one site  are not
directly applicable to the use of the same technology at another
site. Also, much of the available cost data were collected during
demonstration  tests,  and clearly  some of the costs incurred
during  research and development of the technology will not be
incurred again asthetechnology is applied at othersites (Evans,
1990).
  Often cost information for in situ remediation is put in terms of
a cost  per unit of soil treated, and the terms of cost per  cubic
meter of soil treated will be used in this Issue Paper.  However,
the limitation of reporting cost data in this manner must be
understood. Some costs associated with the technology,  for
example, mobilization,  startup, and demobilization,  are  fixed
costs and are not dependent on the volume of soil to be treated.
Other costs,  such as  utilities for steam generation,  are very
dependent on the volume of soil to be treated. Thus, the cost in
terms of per cubic  meter  of soil to be treated are  going  to
decrease as the fixed costs are spread over a greater volume of
soil to be treated. Also, consider the question of depth of the
contaminated zone  to  be  treated.   The deeper the zone  of
contamination, the greaterthe drilling and well installation costs,
but this  may not significantly increase the cost per unit of soil
treated.  However, as the zone of contamination gets deeper,
costs for excavation will increase significantly, and the relative
costs of an ex situ treatment which relies on excavation of the soil
versus an in situ treatment such as steam injection are going to
change.
  The costs associated with the "Detoxifier" demonstration  at
the Annex Terminal Port of Los Angeles have been estimated to
be $330/m3 to $415/m3, based on a volume of 6,824  m3 to be
treated.  The range in costs is related to the percent time on line
that is used to make the estimate. Forty-four percent of the total
costs were labor costs.  Based on the company's operations to
that time with the Detoxifier, they estimate the costs to treat 9,175
m3 of material to be $425/m3 at the treatment rate of 3.8 m3/hour,
$260/m3 at the treatment rate of 7.7 m3/hour, and $165/m3 at a
treatment rate of 15 m3/hour. They expect the costs to reduce as
they gain additional experience and make modifications to their
equipment.   These  numbers show that unit  costs  are very
sensitive to  the treatment rate, which is  determined by site
characteristics, contaminants present, and the clean-up criteria
(EPA, 1991).
  For the steam enhanced remediation project at the  Rainbow
Disposal site, the costs associated with the project were
approximately $4.4 million.  The total amount of soil  that was
considered to have undergone treatment was 72,600 m3, which
gives a cost per  cubic meter of approximately $60.  Labor
accounted for about one-third of the total cost, and utilities
accounted for another large portion. Forthis project, mechanical
problems reduced the on-line factor to about 50 percent, which
drove up the total costs, especially labor costs.  It is reasonable
to expect higher on-line factors at commercial remediation sites.
Cost estimates were developed based on this project using on-
line factors of 75 and 100 percent, and the associated costs were
estimated to be $47 and $38/m3, respectively.
  In the Dynamic Underground Stripping demonstration project,
steam injection heating costs were approximately $2.0/m3 (Udell,
1997), while electrical resistance heating  costs were $6.5/m3
(Newmark, 1992). Overall costs to clean up the gasoline spill site
were $11 million, which works out to a unit cost of about $78/m3
to $92/m3, and they expect the costs to drop by at least one-half
as experience with the process is gained. These costs were also
inflated  due to the fact that  this was a research  project  that
involved a large numberof researchers, and several experimental
techniques were investigated which, while adding to the overall
costs, did little or nothing to  improve the efficiency of the project.
Despite the inflated costs of this project, the total costs  compare
very favorably with estimated costs of $25 million for vacuum
extraction and water table lowering without steam injection, and
$30 million for excavation and backfilling (Yow et al.,  1995).
  Costs of the Dual Auger Rotary Steam Stripping demonstration
project totaled $919,650 and a total of 1566 m3 of soil were
treated,  giving overall treatment costs for this demonstration of
$587/m3. However,significantdowntimewasexperiencedduring
the early part of this demonstration due to equipment problems
and, as mentioned previously, the  rate  of treatment in some
                                                          13

-------
areas was  slowed  significantly due to the undersized vapor
treatment system. With improved on-line times and when higher
treatment rates can be achieved, it is estimated that treatment
costs as low as $88/m3 may be possible.
Conclusions
  Laboratory studies  and  field  demonstrations have
demonstrated the ability of steam injection to effectively recover
volatile and semivolatile contaminants from  the subsurface.
However, in orderto effectively and efficiently apply this process,
it is important to characterize the site adequately to determine
the horizontal and vertical distribution of the contaminant, and
the preferred flow paths forthe injected steam.  This information
is critical to the design of the steam injection and  extraction
system.  Effective operation of the system will  likely include
cyclic operation of steam injection  and vacuum extraction after
steam  breakthrough at the  extraction well has  occurred.
Advantages of steam injection over other remediation techniques
include  the fact that  excavation  is not required,  potential
contaminants are not injected to the subsurface, and potentially
much more rapid remediations are possible. Without a doubt,
the initial capital costs for steam injection are higher than those
for a system that relies on removal of soil gases without heating,
such as vacuum extraction.  However, the accelerated removal
rates can lower the total  cost of cleanup by reducing the time
required forthe remediation, thus reducing the  overall operating
costs (Udell  and Stewart, 1989).   In addition, the  higher
temperatures can increase the amount of semivolatile organics
that are recovered and the removal efficiencies from clay soils by
increasing the volatilization and desorption from soil surfaces. In
systems where the volatilization is limited by a low volatility of the
contaminant  or strong  adsorption  onto a solid phase, the
temperature of the system may actually determine the clean-up
level that can be attained. There will undoubtedly be trade offs
between the  efficiency of  the cleanup and  the cost  of the
treatment process.
Notice
  The U.S.  Environmental Protection Agency through its Office
of Research and Development funded the research  described
here.  It has been subjected to the Agency's  peer  and
administrative review and has  been approved for publication as
an  EPA document.  Mention of trade  names or commercial
products does not constitute endorsement of recommendation
for use.
References
Aines, R., R. Newmark, W.  McConachie, K. Udell, D. Rice, A.
    Ramirez, W. Siegel,  M. Buettner, W. Daily, P. Krauter, E.
    Folsom,  A.  Boegel, D.  Bishop,  Dynamic  underground
    stripping  project, UCRL-JC-109902, Waste  Management
    Symposium, Tucson, Arizona, March 1-5, 1992.
Atkins, P. W., Physical Chemistry, Third Edition, W. H. Freeman
    and Company,  New York,  1986.
Baker,  P.  E., An experimental study of heat flow in  steam
    flooding, Soc. Petrol. Eng. J.,  89-99, March 1969.
Barber, T. E., W. G.  Fisher, and E. A. Wachter,  Characterization
    of the vapor stream  at the Lawrence Livermore Dynamic
    Stripping site  by  differential ultraviolet absorption
    spectroscopy (DUVAS), Dynamic Underground Stripping
    Project: LLNL Gasoline Spill Demonstration Report, UCRL-
    ID-116964, July 1994.
Basel, M. D., and K. S. Udell, Two-dimensional study of steam
    injection into porous media, Multiphase Transport in Porous
    Media, ASME HTD-127, 39-46, 1989.
Basel, M. D., and K. S. Udell, Effect of heterogeneities on the
    shape of condensation fronts in porous media, Heat Transfer
    in Geophysical Media, ASME HTD,  172:63-70, 1991.
Basile, A. J., and G. Smith,  Innovative treatment combination
    rings bell for AT&T, Hazmat World, 52-53, March 1994.
Boberg, T. C., Thermal Methods of Oil Recovery, John Wiley &
    Sons, New York,  1988.
Bursell, C. G., H. J.  Taggart,  and H. A. deMirjian, Thermal
    displacement tests and results Kern  River Field, California,
    Producers Monthly, 18-24, September 1966.
Chu, C., State-of-the-art review of steamflood field projects,  J.
    Petrol. Technol., 1887-1902, October 1985.
Davis, E. L, How heat can enhance in-situ soil  and aquifer
    remediation: Important chemical properties and guidance
    on choosing the appropriate technique, Robert S. Kerr
    Environmental  Research Laboratory, EPA/540/S-97/502,
    April  1997.
Davis, E.  L., and B. K. Lien, Laboratory study on the use of hot
    water to recover light oily wastes from sands, EPA/6QO/R-
    93/021, Roberts. Kerr Environmental Research Laboratory,
    Ada,  Oklahoma, February  1993.
Departmentof Energy, Dualauger rotary steamstrippingPinellas
    plant northeast site, Largo, Florida, Innovative Treatment
    Remediation Demonstration, Costand Performance Report,
    Review Draft, July 21, 1997.
de Percin, P. R., Demonstration of in situ steam and  hot-air
    stripping technology,  J.  Air Waste  Manage. Assoc.,
    41(6):873-877, 1991.
Edmondson, T. A., Effect of temperature on waterflooding, J.  of
    Can.  Petrol. Technol., 236-242, October 1965.
Environmental Protection Agency, Toxic Treatments,  In  situ
    steam/hot-air stripping technology,  Applications Analysis
    Report, EPA/540/A5-90/008, Risk Reduction Engineering
    Laboratory,  Cincinnati, OH, March 1991.
Environmental Protection Ageny, In situ remediation technology
    status report:  Thermal enhancements, Office  of Solid
    Waste and Emergency Response, Technology Innovation
    Office, EPA/542-K-94-009, April 1995a.
Environmental  Protection Agency,  In situ  steam enhanced
    recovery process  Hughes Environmental Systems, Inc.,
    Innovative Technology Evaluation Report, EPA/540/R-94/
    510,  National Risk Management Research  Laboratory,
    Cincinnati, OH, July 1995b.
Evans,  G. M., Estimating innovative technology costs for the
    SITE program, J. Air Waste Manage. Assoc.,  40(7): 1047-
    1051, 1990.
Falta, R.  W., K.  Pruess,  I. Javandel, and P. A. Witherspoon,
    Numerical modeling of steam injection for the removal  of
    nonaqueousphaseliquidsfromthesubsurface, I. Numerical
    formulation, Water Resour. Res., 28(2):433-449, 1992a.
                                                         14

-------
Falta, R. W., K. Pruess,  I. Javandel, and P. A. Witherspoon,
    Numerical  modeling  of steam injection for the removal of
    nonaqueous phase liquids from the subsurface,  II. Code
    validation and application, Water Resour. Res., 28(2):451-
    465, 1992b.
Fan, Y.-H., and K. S.  Udell, An analysis of the vaporization of
    volatile  organic  contaminants from porous  media  by
    conductive heating, Proceedings oftheASME Heat Transfer
    and Fluids Engineering Divisions, HTD-321,715-721,1995.
Farouq All, S. M., and  R.  F.  Meldau, Current  steamflood
    technology, J. Petrol. Technol., 1332-1342, October 1979.
Farrington, P.,  and T. Sword, Heat Enhanced Heavy Fuel Oil
    Recovery Pilot Test, Naval Supply Center Yorktown, Project
    evaluation  report, Groundwater Technology Government
    Services, March 1994.
Gates, C. F., and B. G. Holmes, Thermal Well Completions and
    Operation,   Seventh  World  Petroleum  Congress
    Proceedings,  3:419-429, 1967.
Hadim, A., F. H. Shah, and G. P. Korfiatis, Laboratory studies of
    steam  stripping  of  LNAPL-contaminated  soils,  J. Soil
    Contamin., 2(1):37-58, 1993.
Hall, A. L, and R. W. Bowman, Operation and performance of the
    Slocum thermal recovery project, J. Petrol. Technol., 402-
    408, April 1973.
Herbeck, E. F.,  R. C. Heintz, and J. R. Hastings, Fundamentals
    of tertiary oil recovery Part 8 - Thermal recovery by hot fluid
    injection, Petrol. Eng., 24-34, August 1976.

Hilberts, B., In  Situ Steam Stripping, Assink, J. W.,  Van Den
    Brink, W. J., Eds., Contaminated Soil, Proc. of First Intern.
    TNO Conf.  on Contaminated Soil, Utrecht, The Netherlands,
    pp. 680-687, Nov. 11-15, 1985.

Hunt, J. R., N. Sitar, and K. S. Udell, Nonaqueous phase liquid
    transport and  cleanup,  2.  Experimental studies, Water
    Resour.  Res., 24(8):1259~1267, 1988.
Itamura, M. T., and K. S. Udell, Experimental clean-up of a dense
    nonaqueous phase  liquid  in the  unsaturated zone of a
    porous medium using steam injection, Multiphase Transport
    in Porous Media,  HTD-265, 57-62, 1993.
Itamura, M. T.,  and K. S. Udell, An analysis of optimal cycling
    time and ultimate chlorinated hydrocarbon removal from
    heterogeneous  media using  cyclic steam injection,
    Proceedings  of  the ASME Heat Transfer  and Fluids
    Engineering Divisions, HTD-321, 651-660, 1995.
Johnson, F. S., C. J. Walker, and A. F. Bayazeed, Oilvaporization
    during steamflooding, J. Petrol. Technol.,  731-742, June
    1971.
Keyes, B. R., andG. D. Silcox, Fundamental study of thethermal
    desorption of toluene from montmorillonite  clay particles,
    Environ. Sci. Technol., 28:840-849, 1994.
Langry, K. C., and T. J. Kulp, Effluent stream monitoring at the
    dynamic stripping demonstration using Fourier-transform
    infrared  spectroscopy, Dynamic Underground  Stripping
    Project: LLNL Gasoline Spill Demonstration Report, UCRL-
    ID-116964, July 1994.
Lighty,  J.  S.,  D.  W. Persing, V. A. Cundy, and  D. G.  Linz,
    Characterization of thermal desorption phenomena for the
    cleanup of contaminated soil, Nucl. Chem. Waste Manage.,
    8:225-237, 1988.
Lighty, J. S., G. D. Silcox, D. W. Persing, V. A. Cundy, and D. G.
    Linz,  Fundamentals  for the thermal remediation  of
    contaminated soils.  Particle and bed desorption models,
    Environ. Sci.  Technol., 24(5):750-757, 1990.
Lingineni, S., and  V. K. Dhir, An experimental and theoretical
    study of  remediation  of multicomponent organic
    contaminants  in the  unsaturated  soil by  venting,
    Fundamentals of Heat Transfer in Porous Media, HTD-193,
    99-107, 1992.
Marx, J. W., and  R. H. Langenheim, Reservoir heating by hot
    fluid injection, Trans. AIME, 216:312-315, 1969.
Myhill, M. A., and G. L. Stegemeier, Steam-drive correlation and
    prediction, J. Petrol. Technol., 173-182, February 1978.
Newmark, R. L., Dynamic underground stripping demonstration
    project, UCRL-ID-110064, Interim Engineering Report, U.S.
    DOE Contract No. W-7405-Eng-48, April 1992.
Newmark, R. L., and R. D. Aines, Summary of the LLNL gasoline
    spill demonstration-Dynamic Underground Stripping Project,
    Lawrence Livermore National Laboratory,  Berkeley
    Environmental Restoration Center, UCRL-ID-120416, April
    3, 1995.
Poston, S. W., S. Ysrael, A. K. M. S. Hossain, E. F. Montgomery,
    andH. J. Ramey, Jr., The effect of temperature on irreducible
    water saturation and relative permeability of unconsolidated
    sands, Soc. Petrol. Eng. J.,  171-180, June 1970.

Powers, M. L, C.  J. Dodson, F. Ghassemi, and J. S.  Moore,
    Commercial  application of steamflooding   an  oilfield
    comprising multiple thin sand reservoirs, J. Petrol. Technol.,
    1707-1715, September 1985.

Prats, M., The heat efficiency of thermal recovery processes, J.
    Petrol. Technol., 323-332, March 1969.

Ramirez, A., W. Daily, D. LaBrecque, E. Owen, and D. Chestnut,
    Monitoring an underground  steam injection process using
    electrical  resistance tomography, Water  Resour. Res.,
    29(1):73-87, 1993.

Rubin,  B., and W. L. Buchanan, A general purpose thermal
    model, Soc. Petrol. Eng. J., 25(2):202-214, 1985.
Schumacher, M. M., Enhanced Recovery of Residual and Heavy
    Oils, Second Edition, Noyes Data Corporation,  Park Ridge,
    New Jersey,  1980.
                                                         15

-------
Singhal, A. K.,  Physical model study of inverted seven-spot
    steamfloods in a pool containing conventional heavy oil, J.
    Can. Petrol. Technol., 123-134, September 1980.
Spillette, A. G., Heat transfer during hot fluid injection into an oil
    reservoir,  J.  Can.  Petrol. Technol.,  213-218,  October-
    November 1965.
Stewart,  L D.,  and K. S. Udell, Mechanism of residual  oil
    displacement by steam injection, SPE Res. Eng., 1233-
    1242, November 1988.
Tognotti,  L, M.  Flytzani-Stephanopoulos, A.  F. Sarofim,  H.
    Kopsinis, and M. Stoukides, Study of adsorption-desorption
    of contaminants  on single soil  particles using the
    electrodynamic thermogravimetric analyzer,  Environ. Sci.
    Technol., 25(1):104-109, 1991.
Udell, K. S., Thermally enhanced removal of liquid hydrocarbon
    contaminants from soils and  ground  water,  Dynamic
    Underground  Stripping  Project:  LLNL  Gasoline  Spill
    Demonstration Project, UCRL-ID-116964, July 1994.
Udell, K. S.,  University of  California at  Berkeley,  Personal
    Communication, 1997.
Udell, K. S., M. Itamura, L. Alvarez-Cohen, and M. Hernandez,
    NAS Lemoore JP-5 cleanup  demonstration,  Berkeley
    Environmental Restoration Center, University of California,
    Berkeley, 1994.
Udell, K. S., and R. McCarter, Treatability tests of steam enhanced
    extraction forthe removal of wood treatment chemicals from
    Visalia Pole Yard soils Final Report, Berkeley Environmental
    Restoration Center,  University of  California, Berkeley,
    February 19, 1996.
Udell, K. S., and L. D. Stewart, Jr., Field study of in situ steam
    injection and vacuum extraction  for recovery of volatile
    organic solvents, Department of Mechanical  Engineering,
    University of California, June  1989.
Udell, K. S., and L. D. Stewart, Combined  steam  injection and
    vacuum extraction for  aquifer cleanup, In: Subsurface
    Contamination by Immiscible Fluids,  Weyer, K.  U., ed.,
    Proceedings of the International Conference on Subsurface
    Contamination by Immiscible Fluids, Calgary, Canada, April
    18-20, 1990.
van Lookeren, J.,  Calculation methods for  linear  and radial
    steam flow in oil reservoirs, Soc. Petrol.  Eng. J., 427-439,
    June 1983.
Vaughan, P. J., K. S. Udell, and M. J. Wilt, The effects of steam
    injection of the electrical conductivity of an  unconsolidated
    sand saturated with  a  salt solution,  J. Geophys.  Res.,
    98(B1):509-518, 1993.
Vinsome, P. K.  W., A numerical description  of hot-water and
    steam drives by the finite-difference method, Trans. AIME,
    1-7, 1974.
Weyland, H. V., J. P. Lassiter, and E. J. Veith,  Case history of
    chemical processes associated with the light oil steamflood
    at Elk  Hills, California, Enhanced Oil Recovery,  20-30,
    1991.
Willhite, G. P., Waterflooding, Society of Petroleum Engineers,
    Richardson, TX, 1986.
Willman, B. T.,  V. Valleroy,  G. Runberg, A. Cornelius, and L.
    Powers, Laboratory studies of oil recovery by steam injection,
    Trans. AIME, 681-690, July  1961.
Wilson, J. L, S. H. Conrad, W. R. Mason, W. Peplinski, and E.
    Hagen, Laboratory investigations of residual liquid organics
    from spills,  leaks and the disposal of hazardous wastes in
    ground-water, EPA/600/6-90/004,  Robert  S.  Kerr
    Environmental Research Laboratory, Ada, Oklahoma, April
    1990.
Wu, C. H., A critical review of steamflood mechanisms, Society
    of Petroleum Engineers Paper SPE 6550, 1977.
Yow, Jr., J. L, R.  D. Aines, and R. L. Newmark, Demolishing
    NAPLs,  Civil Engineering, 57-59, August 1995.
Yuan, Z. G., Steam distillation of  liquid hydrocarbon mixtures in
    porous  media, Dissertation,  Mechanical  Engineering
    Department, University of California at Berkeley, 1990.
                                                          16

-------