United States Office of Research EPA/600/R-04/185
Environmental Protection and Development September 2004
Agency Washington, DC 20460
&EPA Filter Fence Design
for Sediment Control at
Construction Sites
-------
EPA/600/R-04/185
September 2004
Filter Fence Design Aid for Sediment
Control at Construction Sites
By
Ellen Stevens, Ph.D.
Assistant Researcher
Billy J. Barfield, P.E., Ph.D.
Professor Emeritus
Department of Agricultural Engineering
Oklahoma State University
Stillwater, Oklahoma
S. L. Britton
Hydraulic Engineer
USDA-ARS Hydraulics Laboratory
Stillwater, Oklahoma
J.S. Hayes
Associate Dean for Environmental Conservation
Clemson University
Clemson, South Carolina
Contract No. 3C-R023-NTEX
Project Officer
Ariamalar Selvakumar
Urban Watershed Management Branch
Water Supply and Water Resources Division
National Risk Management Research Laboratory
Edison, NJ 08837
NATIONAL RISK MANAGEMENT RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OH 45268
-------
Notice
The U.S. Environmental Protection Agency through its Office of Research and Development
partially funded and collaborated in the research described here under Contract Number 3C-
R023-NTEX to Oklahoma State University. The document has been reviewed in accordance with
the U.S. Environmental Protection Agency's peer and administrative policies and approved for
publication. Mention of trade names, commercial products, or design procedures does not
constitute endorsement or recommendation for use.
-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the
Nation's land, air, and water resources. Under a mandate of national environmental laws, the
Agency strives to formulate and implement actions leading to a compatible balance between
human activities and the ability of natural systems to support and nurture life. To meet this
mandate, EPA's research program is providing data and technical support for solving
environmental problems today and building a science knowledge base necessary to manage our
ecological resources wisely, understand how pollutants affect our health, and prevent or reduce
environmental risks in the future.
The National Risk Management Research Laboratory (NRMRL) is the Agency's center for
investigation of technological and management approaches for preventing and reducing risks
from pollution that threaten human health and the environment. The focus of the Laboratory's
research program is on methods and their cost-effectiveness for prevention and control of
pollution to air, land, water, and subsurface resources; protection of water quality in public water
systems; remediation of contaminated sites, sediments and ground water; prevention and control
of indoor air pollution; and restoration of ecosystems. NRMRL collaborates with both public and
private sector partners to foster technologies that reduce the cost of compliance and to anticipate
emerging problems. NRMRL's research provides solutions to environmental problems by:
developing and promoting technologies that protect and improve the environment; advancing
scientific and engineering information to support regulatory and policy decisions; and providing
the technical support and information transfer to ensure implementation of environmental
regulations and strategies at the national, State, and community levels.
This publication has been produced as part of the Laboratory's strategic long-term research plan.
It is published and made available by EPA's Office of Research and Development to assist the
user community and to link researchers with their clients.
Lawrence W. Reiter, Acting Director
National Risk Management Research Laboratory
in
-------
Abstract
The focus of environmental policy and regulation is increasing on water quality issues.
Particularly, there is a more widespread awareness that sediment is one of the most prevalent
pollutants and that the impacts of excess sediment released into lakes and rivers can be as
damaging as those caused by agricultural or industrial chemicals. Due to their nature,
construction sites are typically principal sources of undesirable sediment releases. To make
construction activity easier, sites are generally cleared of all vegetation. The exposed soil is then
made further susceptible to erosion by being disturbed by grading and vehicle traffic. Frequently,
the only action taken to attempt to control sediment releases is the installation of a filter/silt fence.
This approach is not generally successful, for several reasons:
The fence is not installed as recommended by existing guidelines.
The fence is not adequately maintained over time.
The fence is not located for effective control of sediment.
The site is not suitable for a silt fence.
The first two items can best be addressed through public education along with adoption and
enforcement of regulations. The third and fourth items can be addressed through development of
a design aid, which was the objective of this research.
Development of the design aid required the ability to mathematically model the delivery of runoff
and sediment to a silt fence from the drainage area, erosion along the toe, and the behavior of
water impounded behind the fence. Many of the functions in the design aid were adopted from
well-known, established modeling practices. However, existing relationships describing sediment
delivery and concentrated flow erosion are not applicable to the highly disturbed construction site
environment, particularly since it is likely that the soil at these sites is typically excavated and
replaced. Accordingly, these relationships either required adjustment or new relationships had to
be developed. In addition, the hydraulics of flow through and along the silt fences had to be
modeled as there were still gaps in understanding these mechanics.
To develop the additional information needed to complete a silt fence model, a limited series of
flume experiments was completed and a comprehensive series of field-scale tests was conducted.
Sufficient information was obtained for a first-generation model. The purpose of the field tests
was to study erosion along the toe and quantify the amounts of water and sediment delivered to
the fence, flowing along the toe, and flowing through the fence. Primarily, the field data were
used in the model development. In some cases, the existing relationships merely required an
adjustment.
Observations made during a series of construction site visits and during the field experiments
were summarized into a series of recommendations for silt fence siting, installation, and
maintenance. The design algorithms were incorporated into a spreadsheet model wherein the user
can enter site, rainfall, and fabric information, run a hydrologic/hydraulic computation, and then
assess the likelihood of failure and the performance (in terms of sediment trapped) of the silt
fence. The user can vary parameters and see the impact on performance and thereby make the
best possible use of the silt fence on a particular construction site. Finally, the model can also
assist the user in determining when maintenance may be required.
IV
-------
Contents
Notice ii
Foreword iii
Abstract iv
Contents v
List of Tables vii
List of Figures viii
Chapter 1 Introduction 1
Project Description 1
Background 1
Chapter 2 Findings/Conclusions 5
Findings/Conclusions from Field Visits 5
Findings/Conclusions from Simulated Field Testing 6
Chapter 3 Recommendations 8
Siting 8
Installation 9
Maintenance 9
Chapter 4 Experimental Methods - Investigations and Simulated Field Experiments 11
Site Investigations 11
Simulated Field Testing 12
Objectives 12
Field Plot Design 13
Simulated Field Experimental Methods 19
Chapter 5 Results and Discussion - Investigations and Simulated Field Experiments 22
Construction Site Investigation Results 22
Simulated Field Testing Results 23
Results Obtained by Observation 23
Results of Hydrologic Analysis 24
Results Based on Sediment Analysis 26
Analysis of Variance and Trends 29
Chapter 6 Modeling Silt Fence Performance 33
Background 33
Development of Model Components 34
Hydrology Component 34
Hydraulic Component 37
Sediment Component 40
Validation of Model 47
Qualitative Assessment 47
Quantitative Assessment 48
Chapter 7 Design Aid Spreadsheet 54
Background 54
User Instructions 54
-------
Example 55
References 58
VI
-------
List of Tables
Table 4-1. Sites visited in the vicinity of Stillwater, OK 12
Table 4-2. Grid of parameters varied over the simulations 12
Table 4-3. List of simulation dates and parameters 13
Table 4-4. Design parameters for rainfall simulator 17
Table 5-1. Hydrologic parameters calculated from flow data 25
Table 5-2. Sediment results for source area 27
Table 5-3. Sediment results for toe trench and passing through the fence 28
Table 5-4. Analysis of Variance (ANOVA) results 30
Table 5-5. Summary of trends assessment 31
Table 6-1. Assessment of trends in model output 47
Table 7-1. Summary of user input to design aid 55
Table 7-2. Site properties for design aid model 56
Vll
-------
List of Figures
Figure 4-1. Conceptual drawing of the Silt Fence Test Site (SFTS) 14
Figure 4-2. Covered sloped area between samplers and test section of silt fence 15
Figure 4-3. Source area runoff samplers 15
Figure 4-4. Plan and profile of the rainfall simulators 16
Figure 4-5. Sampling locations upslope and downslope of the silt fence 17
Figure 4-6. Pump connection to the simulator 18
Figure 4-7. Schematic silt fence installation 19
Figure 6-1. Wischmeier et al. (1971) nomograph for determining K factor 36
Figure 6-2. Cover factor for use of stockpiled soil 37
Figure 6-3. Triangular channel geometry 38
Figure 6-4. Schematic for four point grid solution matrix 39
Figure 6-5. Comparison of observed and predicted concentrations in toe trench using
equation (6.15) 41
Figure 6-6. Comparison of observed and predicted concentrations by simulation number
(given in Table 4-2) 42
Figure 6-7. Impoundment geometry 45
Figure 6-8. Schematic of flow through silt fence around impoundment 46
Figure 6-9. Model validation - observed vs. predicted average runoff rate from source
(ft3/s) 49
Figure 6-10. Model validation - observed vs. predicted average sediment yield from
source (ton) 49
Figure 6-11. Model validation - observed vs. predicted average discharge in toe trench
(ft3/s) 50
Figure 6-12. Model validation - observed vs. predicted average concentration in toe
trench (mg/L) 51
Figure 6-13. Model validation - observed vs. predicted sediment discharged at end of toe
(lb) 51
Figure 6-14. Model validation - observed vs. predicted net erosion/deposition (lb) 52
Figure 6-15. Model validation - observed vs. predicted concentration of sediment in flow
through fence (mg/L) 53
Figure 7-1. Screen capture of spreadsheet input and output 57
Vlll
-------
Chapter 1
Introduction
Project Description
This report covers the activities completed under Contract # 3C-R023-NTEX, Filter Fence
Design Aid for Sediment Control at Construction Sites. Filter fence and silt fence are used
interchangeably in this report. The objective of the project was to better understand and model
the failure mode of a silt fence as a result of flow along the toe of the fence. Such flow can result
in erosion of the soil that holds the fence's toe in the ground, resulting in eventual undercutting.
The project was divided into three phases as follows:
Phase I:
Create, test, and revise a preliminary design model to assess silt fence installations and guide
the planning of the field testing.
Conduct field assessments of existing silt fence sites.
Phase II:
Evaluate the preliminary design model using data from the site visits.
Use the preliminary design model to determine final design of filter fence and protocol for
field testing.
Develop final procedures for field testing.
Phase III:
Conduct field testing under a range of conditions to clarify the hydraulics, sediment transport,
conditions leading to failure, and overall performance of silt fences.
Combine information from the field testing with previously developed hydrology, hydraulic,
and sedimentation models to develop a model of silt fence performance.
Develop practice recommendations and a design aid based on the model.
Background
Sediment is a pollutant of concern because it is one of the leading causes of stream impairment
across the United States and results in degradation of aquatic life. At a recent EPA invitation-
only conference on sediment control in Cincinnati, OH (EPA, 2002), the lack of effective and
economic technologies for sediment control was identified as a major issue. This paucity of
technologies is a significant problem for all construction and maintenance operations in
-------
residential, commercial, and industrial development, the petroleum industry, highway and other
infrastructure construction, and other activities requiring earth moving.
Currently, a silt fence is the most frequently used structural best management practice (BMP)
technology that does not cause disruption of additional off-site space. As currently constructed,
filter fences consist of a geotextile filter fabric supported by posts and (ideally) anchored along
the toe. Its purpose is to retain sediment from small-disturbed areas by providing a detention time
to allow for sediment deposition (Smolen et al., 1998). Sediment ponds, for example, are also
used, but require the use and disruption of additional land. Although silt fences are widely used,
it has been found, in a recent national study (EPA, 2002), that they are usually ineffective.
Further, on some construction operations, silt fences have been found to cause a concentration of
overland flow, creating worse problems than having no BMP at all.
Although several laboratory studies have proposed that a silt fence can be effective in trapping a
high percentage of sediment, the very few limited field evaluations that have been conducted
indicate that field installations of silt fences trap a very low percentage of sediment (Barrett et al.,
1995). Field inspections (Barfield and Hayes, 1992; 1997) have also found that silt fences were
seldom installed according to standards and specifications, and when actually installed according
to the best current standards, they were still not effective in controlling sediment. Further,
overland flow concentrated by the silt fence can seek the weakest spot on the fence and undercut
the fence or flow through cuts in the fabric. The result is that shallow overland flow coming into
the site is transformed into concentrated flow downslope from the fence, actually increasing the
amount of erosion.
Studies have shown that by removing the surface cover and disturbing the parent soil material,
construction operations increase sediment yield by as much as 10,000 times that of undisturbed
sites (Haan et al., 1994). As this excess sediment moves into streams and waterways it not only
increases the cost of water treatment and reduces reservoir storage capacity through deposition,
but also modifies the stream systems and destroys the habitat of many of our desirable aquatic
species (Smith et al., 1992; EPA, 2001). Ongoing research by the Agricultural Research Service
(ARS) showed that the reduction in species diversity is strongly related to the number of hours
sediment load exceeds 1,000 mg/L, a sediment concentration that is frequently two orders of
magnitude below that in runoff from most construction sites (EPA, 2002). Clearly,
methodologies are needed to reduce sediment loads to levels that maintain habitat and species
diversity. The only method currently available that does not disturb large amounts of additional
landscape is a silt fence, which has not proven to be effective, as will be discussed below.
Laboratory studies of the performance of silt fences using carefully controlled conditions have
cited trapping efficiencies in the range of 40 to 100%, depending on the type of fabric, overflow
rate, and detention time (Barrett et al., 1995; Wyant, 1980; Wishowski et al., 1998; Britton et al.,
2001). Based on these data, the EPA reported in 1993 (EPA, 1993) that a silt fence can have
trapping efficiencies for total suspended solids of 70%, for sand of 80 to 90%, for silt loam of 50
to 80%, and for silty clay loam of 0 to 20%. A recent evaluation of sediment control technologies
conducted by the EPA has not substantiated these claims (EPA, 2002). The results cited in this
study show that field-trapping efficiencies are very low. In fact, Barrett et al. (1995) obtained a
value of 0% trapping, averaged over several samples with a standard error of 26%. Barrett et al.
(1995) speculate that the field tests do not show results similar to lab tests because of: 1)
inadequate fabric splices; 2) sustained failure to correct fence damage resulting from overtopping;
3) large holes in the fabric; 4) under-runs or under-cutting due to erosion of the toe ditch; and 5)
silt fence damage and partially covered by the temporary placement of stockpiles of materials.
-------
Field inspections conducted by Barfield and Hayes (1992; 1997) in which more than 50
construction sites were visited in South Carolina and Kentucky revealed that silt fences were
frequently not installed according to standards and specifications, and further, were frequently
ineffective when actually installed according to standards. In those areas where installations did
meet standards and specifications, lateral flow often occurred along the toe of the fence until
finding the weakest spot on the fence. At that point, it either undercuts the fence, flowed through
cuts in the fabric, or overtopped. Thus, the fence converted shallow-overland flow into
downslope concentrated flow, frequently causing significant, concentrated-flow (gully) erosion.
One recent long-term study has shown promise for the use of silt fences. The US Forest Service
at their Rocky Mountain Research Station (Robichaud and Brown, 2002; Robichaud et al., 2000)
investigated the use of a silt fence as a tool to measure erosion. In this study, they placed the silt
fence across the slope and curled the ends uphill to prevent flow from going around the ends of
the fence. In addition, they buried the toe of the fence upslope from the location of the fabric and
wrapped the fabric around the toe trench to prevent flow over the exposed toe. An average
trapping efficiency of 93% was measured over the season. They cited cleanout and maintenance
as requirements to make the silt fence perform reliably.
Reasons for the poor performance of a traditional silt fence include:
1. Erosion and failure of the toe of the fence from concentrated flow caused by cross contour
installations.
2. Failure to trap fines due to inadequate detention time.
3. Structural problems, including;
a. inadequate strength of the fence fabric resulting in failure from excessive stretching in
the downstream direction, and
b. breaking and overturning of the support post due to inadequate strength and stability of
the footing.
4. Post-installation problems, such as;
a. vandalism as well as destruction by construction equipment, and
b. lack of maintenance.
Problems 1 to 3 can be solved by more effective design as can parts of problem 4. Complete
solution of problem 4 will also require more effective regulation and inspection.
Although current silt fences have a high frequency of failure, their continued use has some
positive aspects, such as being relatively inexpensive, adaptable to a wide variety of sites, and
suitable for implementation without major additional disturbances to the landscape beyond those
required for construction. Thus it seems prudent to evaluate the performance of a silt fence under
a range of controlled conditions leading to procedures for the user community to:
Assess the conditions under which a silt fence can be applied to a particular site.
Develop a design that is specific to that site.
Development of such procedures was the ultimate objective of this project. In the development of
the procedures, the following steps were taken:
A preliminary model was selected and used to design a test facility where rainfall could be
generated on an erosive surface and flow directed toward a silt fence oriented at varying
angles to the contour.
-------
The test facility was constructed and calibrated.
Data were collected on three silt fence fabrics with varying cross contour angles using three
different textured soils.
A final model was developed and tested on data generated at the test facility.
The model was incorporated into a spreadsheet computer program that is available as a
design aid.
Recommendations were made for improvements to silt fence design and installation.
Each of these items will be covered in this report.
-------
Chapter 2
Findings/Conclusions
Findings/Conclusions from Field Visits
The field visits were very informative in pointing out the problems with silt fence usage.
Unfortunately, during the period of this study construction activity was at a low; however, there
were opportunities to observe several of sites with different drainage areas, terrains, ages of
installation, and level of on-going construction activity.
Several of the problems observed can be addressed through better design guidance, such as better
positioning of the fence. On several sites, there was bare soil between the fence and the location
it was supposed to protect. On one site, a fence was placed across a roadside ditch, creating the
potential for street flooding if the fence stands and the water in the ditch back up. On another
site, short lengths offence were erected on a hillside in locations where they would not provide
additional protection, and could potentially worsen erosion by concentrating flows.
Existing guidance documents address the issue of where to position the fence for best protection
and to avoid increasing off-site damage (Smolen et al., 1988). However, it appears that problems
with positioning a silt fence result from a lack of knowledge in the user community about the
guidance available. Better efforts on the part of regulating agencies to inform users about the
guidance available should lead to more effective use of silt fences. A complete solution to the
problem of poor positioning will also require a thorough review of erosion control plans
submitted with applications for building permits and more aggressive inspection and enforcement
once a site is in operation.
Follow-up inspection and enforcement is equally important as a complement to good designs.
For example, one problem consistently observed was failure to anchor the toe in the manner
recommended in almost every set of guidelines evaluated, including the EPA's menu of BMPs
(http://cfpub.epa.gov/npdes/stormwater/menuofbmps/site_30.cfm). A silt fence is frequently sold
pre-attached to the posts, and a practice of taking the loose end (that is supposed to be buried in a
toe trench) and simply laying it flat along the ground and covering it with a thin layer of soil was
observed on a number of sites. While there were locations where this method of installation had
not failed, it is obvious that a heavy rain shortly after installation could easily wash away the
piled up soil and result in no anchoring of the toe. Again, guidance and standards are available,
but it appears that the user community is either unaware or unwilling to use proper methods and
inspection and enforcement of regulations is frequently not pursued. In this case, a lack of
willingness to follow the recommended method is understandable, since it is clearly more costly
to dig a trench, line the trench with the toe, and cover and tamp the backfill than it is to throw soil
on a flap of fabric. For now, better education, inspection, and enforcement will improve this
problem. Ultimately, however, contractors will only be willing to anchor the toe correctly if there
is a mechanized or otherwise quick and cost-effective means of doing so. Supporting
development of such a system and promoting its use should be a priority for regulatory agencies.
-------
Fence failures due to damaged posts and fabric were also observed, which is largely a
maintenance issue. However, use of the design aid developed as part of this project can help
users to determine if their fence will be subjected to depths or volumes of water that are likely to
be damaging. In general, a freeboard of 0.5 ft offence above the predicted depth of flow or
impounding is recommended, and the design aid can be used to predict the peak depth.
Many of the problems observed during the site visits, particularly failure to repair broken or
overturned posts or damaged fence, are primarily solved through better inspection and
enforcement. For remote areas that may not be visited too often, the design aid can be used to
estimate the accumulated depth of sediment resulting from atypical storm and provide an
indication as to how often the site should be checked for problems. If there is a nearby national
weather service rain gage or some local weather station to provide inches of rainfall, each storm
can be inputted into the design aid after it occurs and a running tabulation of the predicted
sediment accumulation created.
Findings/Conclusions from Simulated Field Testing
Failure of the toe trench was observed in eight simulations. The credibility of the soil was a
major factor in this, with six of the eight failures occurring with the loam soil. The secondary
factor was the slope, with the other two failures occurring with red clay at a 13.5% slope. The
volume of soil in the toe trench was approximately 10 ft3. Almost all the failures occurred when
the net erosion, i.e., discharge from the toe trench in excess of the incoming sediment load,
reached 25% of that toe trench volume.
Half of the simulations had net erosion and half had net deposition. Again, the soil type and slope
along the toe were the controlling factors. Net deposition occurred for all soil types at the 1%
slope. Net deposition is beneficial in that it helps protect against toe failure due to scour and
represents sediment that is prevented from leaving the site.
Regardless of whether there was net erosion or deposition, there was always significant sediment
discharge at the downstream end of the toe. This sediment needs to be prevented from flowing
off-site. One way to contain this sediment is to extend the silt fence upslope at the ends of the
toe. It is desirable to have an extension at both ends, unless the slope along the toe is relatively
steep. Based on several design aid simulations, it appears that an extension of 10 to 20 ft will be
sufficient in most cases. The actual length required is determined from the peak depth and the
slopes along the toe and toward the fence.
Soil type was also a factor in net deposition. With the very erosion-resistant black clay, there was
net deposition in five or six tests, with a small amount (about 0.25 ft3) of net erosion during one
steep slope test. For sandy loam and loam, there was net erosion for both the moderate and steep
slopes.
In almost all the simulations, a first-flush effect of concentration peaking during the first 15 to 20
min of runoff was observed. Therefore, the most downstream damage is apt to occur
immediately following the initiation of runoff. This points out again the importance of not
discharging the flow along the toe trench directly into a stream or other sensitive area. If the area
for an impoundment or sediment trap is limited, means of capturing, at a minimum, the first 15 to
20 min of discharge should be developed.
-------
The accumulation of sediment in sags in the fence can be beneficial up to a point by creating flat
areas where more deposition is likely. The accumulation of sediment will eventually become a
problem either by causing collapse of the fence under the loading or reducing the height so the
overtopping becomes likely. The design aid can help in determining when there is too much
sediment and when clean-out is needed. During the field tests, an apparent equilibrium was
observed with incoming and discharged sediment loads approaching each other. The design aid
can be used to determine if the soil/slope combination is one that will result in equilibrium under
a given rainfall, thereby making the duration of rainfall much less of an issue.
The field test site usually became saturated and depressions filled in a short time - typically less
than 15 min - giving a constant rate of runoff, usually slightly less than the rate of rainfall. This
indicates that the National Resources Conservation Service (NRCS) curve number method for
generating runoff is valid for construction sites, as the volume of runoff converges to a constant
value under an input of steady rainfall.
In summary, a number of the observations and results from the construction site visits and
simulated field studies confirmed the validity of the recommendations contained in the numerous
guidance documents.
-------
Chapter 3
Recommendations
Observations made at field visits, plus the data and observations collected in the field and flume
testing have contributed greatly to our understanding of how silt fences functions under different
conditions. That data, combined with information gleaned from literature and personal contacts
with other researchers and agencies indicate that a properly installed and maintained silt fence can
significantly reduce the amount of sediment discharged from a disturbed area such as a
construction site. However, this data and literature also indicates that a silt fence is more often
than not installed in a manner that renders it ineffective. Recommendations made earlier by EPA
were based on laboratory data (EPA, 1993) since field observations were not available; however,
this study plus limited additional studies (Barrett et al., 1995) indicate that the laboratory studies
were not representative of actual practice in the field. The results of the study in this report, plus
one field study of the use of a silt fence as a technique to measure sediment yield (Robichaud and
Brown, 2002), indicate that currently available silt fences can be installed in a manner to provide
effective sediment control technology. To make that happen, however, it is critical that the fence
be properly cited, installed, and maintained. In addition, some changes are needed in guidelines
typically used by most regulatory agencies. Therefore, the following procedures with respect to
siting, installation, and maintenance are recommended for addition to current guidelines.
Siting
Restrict the drainage area so that the predicted depth of water at the fence under the design
rainfall does not go above 0.5 ft. The design aid spreadsheet which is described in Chapter 7
can be used to develop the depth prediction. Since sediment is likely to accumulate behind
the fence, restricting the depth will allow for some sediment build-up before overtopping
becomes a problem.
The silt fence should be located on the contour as closely as possible. Since this is frequently
not possible due to site limitations, the combination of cross contour angles and length of silt
fence segment should be limited to prevent erosion of the toe of the fence and undercutting.
The design aid in Chapter 7 can be used to determine the length that will cause erosion on a
site specific basis for a selected design storm.
Silt fences should always be installed in a bowl shape with limits to the length of individual
segments based on the ability of the installation to store the design storm without
overtopping. When installed with the bottom section parallel to the contour, both ends should
have an upslope distance sufficient to store the design storm without having flow discharge
around the upslope end. When installed with a cross contour slope, it should be assumed that
the design storm will be stored in the catchment formed by the upslope extension at the lower
end of the silt fence segment. When calculating the flow through the fence, the flow rate
through the fence should account for the impact of plugging. The design aid in Chapter 7 can
be used to determine the length of the upslope extension and the impacts of plugging.
The required trapping efficiency could be determined by the design aid in Chapter 7 to assure
that the required trapping efficiency will be met, particularly when fine textured soils are
-------
involved. Following design criteria will not always assure that the trapping efficiency
standards will be met.
Installation
To avoid failure of the silt fence, measures should be taken such that the toe does not erode to
a level that causes undercutting. To assist in preventing erosion of the toe, the fence should
be installed so that the fabric lines the ditch excavated for the toe and the ditch is filled with
soil and compacted to a density necessary to prevent erosion of the toe. In order to further
protect the toe trench, the fence fabric that becomes the vertical portion of the fence should
exit the trench upslope and be wrapped over the toe trench before being attached to the posts
which are installed at the downslope edge of the trench.
The depth of the toe trench should be sufficient to prevent undercutting from the design
storm; in general, guidelines and experience indicate a minimum of 6 in. burial depth. When
trenching machines are used, tests should be run with the equipment to ensure that the fence
will not be pulled from the ground when soil is saturated, under full impoundment volume of
stored water, and post spacing equal to or greater than specified in the installation plans.
Also, tests should show that the fence will not fail due to undercutting with the machine
installed systems when used on slopes equal to or greater than those in the installation plans.
Posts should be installed with sufficient strength to resist breaking from the forces of
impounded water and deposited soil during the design storm. Although the fence may be
designed for trapping efficiency in a given design storm, when storms greater than the design
storm occur, more water may be impounded than would occur in the design storm; therefore,
the posts should be designed with sufficient strength to withstand full impoundment. In
addition, field observations of silt fence installations indicate that construction operations
often result in damage to the small posts that are commonly pre-attached to fabric in the
manufacturing process.
To be effective under the high moisture conditions that occur during stormwater events, the
posts should have sufficient bearing surface to remain erect when the soil is saturated and the
storage volume behind the fence is full. In cases where the bearing strength of soil is
insufficient, additional bearing surface should be included by the use of fins, or a different
post geometry and burial depth.
Excessive stretching offence fabric can lead to failure; therefore, a combination of small
spacing between posts and reinforcing offence with either a high strength geotextile grid or
metal web fence backing should be used at all installations.
Maintenance
Fences should be inspected after significant rainfall events (typically 0.5 in. or more) for
significant deposition of sediment behind the fence, undercutting, damaged posts, areas
where the fence has overtopped, and downslope damage from locations offence failure. The
design aid can be used to predict the amount of deposition expected for specific rainfall
events, giving an indication of the amount of storm activity that can occur before
maintenance is needed. Cleaning out of accumulated sediment is recommended whenever the
height offence above the deposition is less than 0.75 ft. This assumes that the area/depth
recommendation in the section on siting is followed.
In sandy or loamy soils, failure by toe erosion is also likely. The toe should be inspected
after heavy rains. The design aid can be used to predict if a specific level of rainfall intensity
will cause failure of the toe.
-------
Preventive maintenance checks should also be made on a regular basis during periods without
significant rainfall events (at least weekly) for damage from construction operations and
vandalism. Repairs should be made as necessary, including but not limited to replacing fence
segments, patching tears and cuts, reinstalling posts that are damaged, reburial of fabric in the
toe trench, and cleanout of deposited sediment.
10
-------
Chapter 4
Experimental Methods - Investigations and Simulated Field
Experiments
Techniques and protocols for the construction site visits and the simulated field experiments are
summarized in this chapter. Information included in this chapter will provide end-users with a
basis for determining if the technique or recommendation presented is applicable to a specific
design situation. Details and development of the mathematical model and design aid are given in
Chapter 6.
Site Investigations
Construction sites in Oklahoma and South Carolina were visited to obtain information about how
silt fence installation practices were actually followed, to observe any problems, and to determine
from visual inspection, if possible, whether the silt fence resulted in a decrease in the sediment
leaving the site. Data from selected sites were used to evaluate the ability of the routines in the
preliminary design model to route the water and sediment and predict erosion and deposition in
the context of flow toward and along a silt fence.
The sites in Oklahoma were chosen to represent the widest variety of conditions possible in terms
of size, terrain, soil, and age and condition of installation. A mixture of well and poorly
performing installations was also considered desirable. Unfortunately, construction activity in the
Stillwater area was down during the original period allotted for the visits, April to June 2003, and
only three appropriate sites were identified. It was necessary to wait until late 2003 to complete
the visits, and the number of active sites where a silt fence was in use was still very limited.
Three additional suitable sites in varying stages of activity were located: a newly-installed fence
on a recently graded area, a site with intense building activity, and a mature site where the fence
was ready for removal.
For each site, as much history as possible was obtained from the owner or contractor. During site
visits, to the extent possible, a prescribed series of observations and measurements was made. At
times, climate, terrain, or vegetation conditions made it necessary to skip some observations or
measurements.
The observations planned for each site included: type of fabric and properties; anchoring of both
toe and posts; fraction of vegetative cover on contributing area, type of vegetation, and fraction of
paved surface; and location of failure points with estimates of causes of failure. To generate the
data needed to evaluate the preliminary design model, a survey was completed to determine
contributing drainage area and slope, slope of the toe of the fence, and angle between contour and
toe offence. The height of the fence, and spacing of posts were also recorded. Soil samples
were gathered for grain size analysis, and digital photographs were taken at each site.
11
-------
During visits in late 2003, the measurements required only for the preliminary design model
assessment were skipped, since that assessment was complete and the field plot had been
constructed by that time. Table 4-1 gives a description of the construction activity and features of
the sites.
Table 4-1. Sites visited in the vicinity of Stillwater, OK
Site
Number
1
2
3
4
5
6
7
Description
Construction of a walking trail -very long
length offence
Construction of a walking trail - large, steep
drainage area
New high school - steep drainage area with
sandy soil
Same as site 2 - revisited because heavy
rains caused failure
Single commercial building - newly installed
fence
Commercial office complex -very active
site
Residential subdivision - construction nearly
completed
Location
Stillwater, OK - near Albertson's
Stillwater, OK - near Fazoli's
Perkins, OK
Stillwater, OK - near Fazoli's
Stillwater, OK - Miller and Main
Stillwater, OK -Western Rd.
Stillwater, OK - Lakeview Rd.
Visit
Date
03/21/03
03/26/03
04/09/03
04/20/03
10/07/03
12/30/03
12/30/03
Simulated Field Testing
Objectives
A series of tests with controlled conditions was conducted at a specially-designed silt fence
testing site located at the USDA-ARS Hydraulic Laboratory. The purposes of the tests were to
(1) supplement the information obtained from the field visits and the literature search with an in-
depth examination of silt fence performance; and (2) obtain numerical data to develop and
validate the final mathematical model.
Parameters that were varied from test to test are summarized in Table 4-2. Factorial experimental
design was used, with all combinations being tested. Dates and parameters for all simulations are
given in Table 4-3.
Table 4-2. Grid of parameters varied over the simulations
Soil Type
Silty clay (sand-15%, silt-44%, clay-41%)
Loam (sand-53%, silt-32%, clay-15%)
Sandy loam (sand-64%, silt-17%, clay-20%)
Slope Along Toe
Steep - approx 13.5 %
Moderate - approx 7 %
Flat- approx 1%
Fabric
Nilex 2127 (Fabric A)
Nilex 21 30 (Fabric C)
12
-------
Table 4-3. List of simulation dates and parameters
Simulation
Date
12/18/2003
1/15/2004
3/2/2004
3/10/2004
3/19/2004
3/24/2004
4/20/2004
4/22/2004
4/28/2004
5/4/2004
5/7/2004
5/11/2004
5/18/2004
5/21/2004
5/25/2004
5/27/2004
6/2/2004
6/4/2004
Soil
Type
Loam
Loam
Loam
Loam
Loam
Loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Silty clay
Silty clay
Silty clay
Silty clay
Silty clay
Silty clay
Fabric
Type
C
A
C
A
A
C
C
A
A
C
C
A
A
C
C
A
A
C
Slope Along Toe
%
Flat
Flat
Moderate
Moderate
Steep
Steep
Steep
Steep
Flat
Flat
Moderate
Moderate
Moderate
Moderate
Steep
Steep
Flat
Flat
Monitoring had to be accommodated in the field plot design in order to determine runoff rates and
sediment yield (1) at the edge of the silt fence's source area, (2) at the downstream end of the toe
of the silt fence, and (3) flow passing through the fence. The experimental methods had to
include provision for obtaining sufficient data to validate the mathematical model.
This section provides an overview of the field plot design and experimental methods and results
are discussed in Chapter 5.
Field Plot Design
In order to provide access to a wide variety of test conditions, a Silt Fence Test Site (SFTS) was
constructed at the USDA-ARS Water Conservation Structures Laboratory in Stillwater, OK. The
need for access to water and electricity was the primary constraint in site selection. A secondary
consideration was to have a terrain that was compatible with the proposed field plot
configuration, as shown in the original conceptual drawing (Figure 4-1). Accordingly, a hillslope
near the large concrete flume was selected.
13
-------
Figure 4-1. Conceptual drawing of the Silt Fence Test Site (SFTS).
Note: Runoff and sediment from the source plot sampled at A, and discharge through the fence is sampled
atB.
Since a rainfall simulator would have to be constructed, limitations to the size of the source area
for sediment and runoff had to be imposed. An area of 20 ft upslope by 40 ft along the fence was
selected as being small enough to be manageable for purposes of constructing a rainfall simulator
yet large enough to represent a large number of field conditions. A 5% slope was selected for the
source area. The covered, sloped area between the samplers and fence (Figure 4-2) needed to be
as steep as possible so that the slope along the toe could vary. However, it also needed to be flat
enough that personnel and equipment could work safely. Balancing these concerns and the need
to minimize the amount of fill required, a 20 ft long, 3:1 slope was selected. This would allow
slopes along the toe to vary from zero to almost 14%.
Source area runoff samplers were installed as shown in Figure 4-3 to briefly sample runoff water
and sediment from the source area and create minimal disturbance to the flow as it progressed
toward the silt fence. To further minimize disruption of the flow, the samplers would operate in
alternating groups, with only half of the samplers open at any one time.
14
-------
Silt Fence and Downslope Sampling
Flow Path When Not Sampling
Figure 4-2. Covered sloped area between samplers and test section of silt fence
Figure 4-3. Source area runoff samplers
For the rainfall simulator to adequately cover the 20 by 40 ft area, it required four rows of
nozzles, seven per row. A central main supply line provided water to the four rows, with three
nozzles per row on one side of the main and four nozzles on the other side of the main. Since the
runoff plot had a 5% slope, the simulator had to be built on a slant, to make all rows of nozzles 10
ft off the ground. Plan and profile of the simulator are shown in Figure 4-4.
15
-------
o
o
o
o
o
-o
o
o
7 at 5.71 ft
2.8
ft
4 at
5ft
10ft
4 at 5 ft
10ft
5% slope
Figure 4-4. Plan and profile of the rainfall simulators
It was desirable to have a simulator that could be programmed to deliver between 1 and 3 in./h.
The nozzles were grouped in rows, with the first row (row closest to the samplers) and third row
operating together and the second and fourth rows also operating together. A controller was
programmed to pulse the nozzle groups at intervals to achieve the desired rainfall intensity. The
final rainfall simulator design parameters are summarized in Table 4-4.
Discharge along the upslope side of the toe of the fence was collected at the end of the fence into
a sheet metal trough with a spout as shown in Figure 4-5. A similar trough with spout was
attached to the end of a triangular trough mounted on the downslope side of the fence, also shown
in Figure 4-5. A sample pit, deep and wide enough for a technician to maneuver a bucket under a
spout and deep enough to hold the bucket in place at the end of the spout, was excavated at the
edge of the plot. The walls of the sample pit, which were approximately 3 ft deep, were
supported with metal plates.
16
-------
Table 4-4. Design parameters for rainfall simulator
Parameter
Nozzle type
Height above ground
Minimum spacing
Operating pressure
Operating discharge
Unit
Spray Systems Fulljet: 1/2 HH-SS 30WSQ
10ft
6ft
4 Ib/in.2
2 gal/min
Q down slope
Q through
fence
Side View
Sample pit to position
bucket at end of
trough
Fence and posts
2-in. high walls
around ends of
troughs to^^
force water
into spouts
Sample
pit wall
6-in. high walls
to separate
troughs
Top View
Figure 4-5. Sampling locations upslope and downslope of the silt fence
A portable pump mounted on a skid was selected to deliver the water to the simulator. The pump
intake was connected to the water source with a 2.5-in. hose, and the pump discharged into a 2.5-
in. hose connected to the simulator main. The pump motor was electric and could be plugged
into the outlet on the power pole at the site. A back-up pump was available if the pump should
break down during a simulation. Figure 4-6 shows the pump connection to the simulator. To
monitor the pressure, a pressure gage was mounted on the main next to the front row of nozzles.
17
-------
Figure 4-6. Pump connection to the simulator
Light-gauge sheet metal with ribs spaced at 1-ft intervals was used to cover the soil between the
front plot wall and the silt fence. The ribs would function to keep the flow distributed along the
plot and also minimize warping and deforming of the sheet metal under extreme temperatures.
The toe of the silt fence was buried in the manner recommended in the EPA menu of BMPs
(http://cfpub.epa.gov/npdes/stormwater/menuofbmps/site_30.cfm). A trench 0.5 ft deep by 0.5 ft
wide was excavated and lined with the toe of the fence. The trench was then backfilled with the
same material as was present on the source area and compacted by hand tamping. This method of
compaction was considered the most representative of what occurs at construction sites. Figure
4-7 is a schematic of the installation.
18
-------
Silt fence with toe
curled around in
trench
0.5ft
Source area soil
compacted in
trench
Sheet metal on 3:1 slope
0.5ft
Figure 4-7. Schematic silt fence installation
Simulated Field Experimental Methods
The following sections describe the data collection activities and information that was measured
and/or derived based on the sampling. The laboratory analyses conducted and methods used are
also described.
Field Data Collection
Field data collection included activities before the start of the simulation, at intervals during the
simulation, and at its conclusion. Data collection activities before each rainfall consisted of
collection of soil samples for bulk density and surveying a pre-test profile of the toe trench. The
samples for bulk density were collected using the drive cylinder method. A total station was used
to complete the survey. Elevations were tied to a benchmark located on the concrete foundation
for the storm sewer grate. For each profile, an angle plus horizontal and vertical distances were
recorded at the edge of the sheet metal, at the center of the toe trench, and at the silt fence. These
measurements were taken at every fence post.
The following sampling and data collection activities took place while the rainfall simulator was
operating:
Samples for flow rate and sediment concentration
o Edge of source area
o Flow along toe of fence
o Flow through fence
Rainfall and wind gauge readings
Video record of flow in toe trench
Visual observations of flow, erosion, etc.
In each test, sampling for flow rate and sediment concentration was initiated as soon as there was
sufficient runoff for an adequate sample. To compute flow rates, the time required to collect each
19
-------
sample was recorded. Samples were collected in pre-weighed containers - either bottles or
buckets, depending on flow rate - and the samples plus containers were weighed to find the
weight and volume collected. When discharge rates were relatively slow, samples were collected
in 1-L bottles. These samples were taken directly to the lab for weighing and sediment
concentration analysis. With higher flow rates, samples were taken in 5-gal buckets, weighed in
the field, and grab samples were taken in 1-L bottles for laboratory analyses.
Two rain gauges were placed in the source area, roughly centered in each half. These were the
small plastic rain gauges that are read by noting the level of the water on the scale printed on the
side of the container. The rain gauges were read at intervals of 10 to 15 min (depending on other
responsibilities).
A video record was made of the flow in the toe trench by means of video cameras set on tripods
at each end of the trench. In addition, the field supervisor observed flow in the toe trench and
through the fence and noted any significant or unusual occurrences, such as the time at which
flow ceased in the collection trough downstream of the fence. Scour and deposition in the toe
trench were monitored and significant events such as the times and locations of complete scour
down to the silt fence were recorded. The observations are summarized in Chapter 5. In
addition, many of the events of the simulation were recorded with a still (digital) camera.
Laboratory Analyses
Laboratory analysis was performed to determine the sediment concentration, soil bulk density,
and particle size of the eroded sediments. Total suspended solids (TSS) testing was performed
using a Syringe Filter Method based on EPA method 160.2, "Standard Procedure to Determine
TSS." The procedure given in ASTM D3977 - 97 was followed to perform the suspended
sediment concentration (SSC) test. Soil bulk density was determined using the Drive Cylinder
Method (ASTM D2937). For particle size, the larger size fractions were determined using a wet
sieve apparatus. The small size fractions passing the wet sieve were analyzed using a Microscan
II x-ray particle analyzer. Laboratory analysis results are discussed in Chapter 5.
Data Reduction
Field measurements and laboratory analyses were used to generate a variety of hydrologic and
sediment results to support the conclusions and aid in the modeling effort.
Flow hydrographs
Sample collection times and volumes obtained were used to develop the time series of discharge
at the edge of the source area, along the toe of the fence, and passing through the fence. Linear
interpolation was used to estimate the discharges between the discrete data points.
Rainfall data
Rain gauge readings were used to determine the cumulative depth vs. time relationship for each
simulation. Because the simulator was programmed for a constant intensity, these relationships
were all very close to linear. Least squares linear regression was used to estimate the rainfall
depths between the discrete data points.
Cumulative flow volumes and average flow rates
The volume discharged between two sample collection times was estimated by finding the area
under the flow hydrograph between those two times and a cumulative volume vs. time
relationship was developed. This relationship was also very close to linear because the runoff
rate converged to a constant value soon after the source area was saturated. The slope of that
20
-------
linear relationship (estimated by least squares regression) was used to estimate the average
discharge rate. These volumes and discharge rates were computed for all three sampling points.
Curve number
The cumulative rainfall and flow volume estimates were used to estimate the NRCS curve
number for each simulation. The runoff volume from the source area was expressed in units of
inches of depth and a value of curve number that minimized the sum of squared errors between
predicted and estimated values of runoff depth was determined.
Sedigraph or sediment flow rate in Ib/s vs. time
Each flow rate sample had an associated sediment sample, which was analyzed for concentration.
The sedigraph ordinates were obtained as the product of the concentration (converted to lb/ft3)
times the flow rate (ft3/s). The sedigraph was then used in the same way as the hydrograph to
estimate the weight of sediment discharged between sampling intervals. Sedigraphs were
computed for all three sampling points.
Cumulative weight of sediment
A cumulative weight of sediment discharged vs. time relationship was developed for all three
sampling points. The average sediment flow rate over the simulation (Ib/s) was estimated by
finding the slope of that relationship.
Average sediment concentration
Average sediment concentration over the simulation period was computed by dividing the
estimated total weight of sediment by the estimated total volume of water and converting to
mg/L. This was computed for all three sampling points.
21
-------
Chapter 5
Results and Discussion - Investigations
and Simulated Field Experiments
Results from the construction site visits and the simulated field experiments are summarized and
discussed in this chapter. These results, along with information obtained from the literature, were
used in development of the mathematical model, design aid, and BMP recommendations. The
majority of the literature citations were included in Chapter 1 and details of the development of
the mathematical model and design aid along with testing the model with field laboratory data are
given in Chapter 6.
Construction Site Investigation Results
Several problems were observed on a number of construction sites visited. One was a failure to
anchor the toe in the recommended manner, particularly where it appeared that the fabric was
purchased pre-attached to the posts. In several cases, the loose fabric that was supposed to be
anchored in the toe trench was simply stretched out flat upslope of the fence and covered with
soil. At those sites visited, the main problem observed with this method of installation was
bulging of the fence from placing the soil against the fabric. In some locations, there was excess
soil behind the fence, reducing the vertical extent of the fabric and making overtopping by water
or accelerated filling with sediment more likely.
Significant sediment accumulation was observed at Site 2 (refer to Table 4-1), where at one
location there was less than 0.5 ft of fabric above the accumulated sediment. The site was re-
visited after about 2.5 in. of rain occurred; the silt fence was completely covered with deposited
sediment.
Lack of maintenance was another observation. The extent of this ranged from failure to repair
localized damage from broken posts or tears in the fabric to failure to repair the total collapse of
the fence to failure to remove deposited sediment.
Improper placement was also observed at some sites. At several sites, disturbed soil was located
between the fence and the area that was supposed to be protected. At one site, a fence was placed
across a drainage swale.
There was also evidence at some of the sites that the silt fence did function to reduce the amount
of sediment transported off-site, although the magnitude of this reduction could not be estimated
from the observations and measurements. Evidence to support that the silt fence was functional
included observation of deposited sediment upslope of the fence, absence of rills or other
evidence of concentrated flow downslope of the fence, and whether or not the fence was intact
and erect.
22
-------
Simulated Field Testing Results
The series of 18 tests conducted showed that the silt fence test site yielded a wide range of useful
information about silt fence hydraulics and overall performance. Each test was a controlled-
conditions test, under simulated rainfall, of commonly-used commercial grades of silt fence
installed to emulate construction site conditions. Field-scale data including runoff rate, sediment
yield, flow and sediment transport along the toe, and sediment transport through the fence were
collected. The tests were structured so that the impact of changes in slope along the toe, soil type,
and fabric type could be assessed.
Insight and data regarding the performance of a silt fence was obtained through visual
observation and photography and through field measurements and laboratory analyses of water
and sediment samples. The following sections discuss the results and present an analysis of
variance and trends.
Results Obtained by Observation
Although informal, a very informative component of the field testing was the record of visual
observations made by the field supervisor and crew. Several interesting phenomena were
observed through this process.
The "first-flush" effect
Once runoff was established, it appeared that the discharge exiting the toe trench at the
downstream end had greater turbidity during the initial phase of the simulation. This was
considered reasonable and was attributed to the fact that there was loose material in the trench
that would be easily dislodged.
This observation is confirmed by analysis of water samples taken from the toe trench. For
practically all simulations, the concentration would peak early in the simulation and then decrease
and level off. There were a couple of exceptions to this, and the observations made at the time do
not suggest a reason.
Accumulation of sediment behind the fence
For tests where there was no erosion, sediment accumulated along the toe, and significant
deformation of the fence occurred. This was similar to the upslope deposition observed during
construction site visits.
Two impacts were associated with this deposition. One was that it created a location where flow
along the toe would tend to slow down, leading to additional deposition occurring. In the short
run, this would help the fence to be more effective at trapping sediment. In the long run, the
accumulated sediment itself can be a source of failure, particularly if the fence is not maintained
regularly.
The other impact of this had more to do with how measurements were made rather than with how
the silt fence functions in the field. The collection trough on the downstream side of the fence
was mounted along the toe of the fence and flush with the ground surface. When bulging of the
fence occurred, it could partially or completely block the trough. For that reason, it appears that
the sampled flow passing through the fence in these cases may not represent the discharge that
would be seen under field conditions. However, the concentrations recorded should be
representative, since the flows that did traverse the trough would not likely pick up additional
sediment. Also, when flow was not blocked in the trough, it appeared to be free-running with no
unusual deposition that would distort the concentration data.
23
-------
Attainment of equilibrium
After a period of time, typically in the range of 20 to 30 min, the scour and deposition appeared to
reach equilibrium. This was observed by noting the depth of scour along the walls of the toe
trench and the presence and movement of bed features such as head cuts, and by probing the soil
in the toe trench with a pointed stake painted with alternating 0.1-ft rings to observe the depth to
the silt fence lining the toe trench.
When equilibrium was observed, the discharge in the toe trench was obviously continuing to
transport sediment, but it appeared that a point had been reached where the amount of sediment
coming in from the source area was equal to the amount of sediment exiting. This was confirmed
by examination of plots of cumulative weight of sediment leaving the source area and exiting the
toe trench at the downstream end. In a number of simulations, there would be a point after which
the plots of cumulative sediment yields were parallel with each other, indicating that mass flow
rates of sediment into and out of the toe trench were roughly equal.
After the first two or three simulations, it was confirmed by looking at the data that the
equilibrium condition was, in fact, occurring so attainment of equilibrium was used as a criterion
for stopping the rainfall. Typically, once equilibrium was observed, the simulation would
continue for 15 to 20 min. and then shut down.
An equilibrium condition was not attained in all simulations, depending on the soil type and
slope. With the highly erodible sandy loam soil, the toe trench failed at all slopes. The
simulation was stopped once there was significant progression of failure. However, it appeared
the toe trench would continue to scour until the silt fence was exposed for its entire length. The
same was true with loam when the slope along the toe was steep (approximately 13.5%).
Results of Hydrologic Analysis
Computations described in Chapter 4 were completed for each simulation, and the results are
summarized in Table 5-1. Once the flow data were all processed and assembled, several
phenomena which were not observable during the simulation became apparent.
The rainfall gauge readings indicated that the simulator was operating at more than 2 in./h, to up
to about 2.5 in./h. This was not considered a problem since the 2 in./h rate was an arbitrary
selection and the hydraulic components of the test plot (samplers, collection troughs, etc.) were
sized for rainfall rates of at least 3 in./h.
Runoff from the source area was more variable than rainfall, particularly early in the simulation.
This is partly due to the disturbed nature of the source area, which would not achieve saturation
uniformly over the area. Also, irregularities in the surface lead to minor ponding and then release
from the ponding. Since this is what would be expected from a field construction site, it was not
considered a problem. Typically, the runoff rate converged to a constant value once the plot was
fully saturated, all depressions were filled, and all areas of the plot were contributing to runoff.
24
-------
Table 5-1. Hydrologic parameters calculated from flow data
Simulation
date
12/18/2003
1/15/2004
3/2/2004
3/10/2004
3/19/2004
3/24/2004
4/20/2004
4/22/2004
4/28/2004
5/4/2004
5/7/2004
5/11/2004
5/18/2004
5/21/2004
5/25/2004
5/27/2004
6/2/2004
6/4/2004
Soil
type
Loam
Loam
Loam
Loam
Loam
Loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Silty clay
Silty clay
Silty clay
Silty clay
Silty clay
Silty clay
Fabric
type
2130
2127
2130
2127
2127
2130
2130
2127
2127
2130
2130
2127
2127
2130
2130
2127
2127
2130
Slope
%
1
1
7.5
7.5
13.5
13.5
13.5
13.5
1
1
7.5
7.5
7.5
7.5
13.5
13.5
1
1
Curve
No
NA1
NA1
99.2
97.0
95.4
99.2
96.8
99.3
93.6
94.2
91.9
89.1
86.1
87.3
85.4
89.2
88.2
92.9
Rf avg
in./h
1.64
1.85
2.85
2.13
2.51
2.85
2.41
2.21
2.76
2.64
2.42
2.52
2.45
2.43
2.62
2.55
2.28
2.62
RO avg
in./h
2.19
3.07
2.43
1.78
2.09
2.43
2.13
1.79
2.11
1.99
1.79
1.43
1.86
1.60
1.52
1.62
1.17
1.60
RO start
time, s
245
978
500.21
1072.05
90.42
509.07
53.74
645.37
377.84
345.12
178.98
453.58
665.67
940.31
394.55
400.08
439.27
1852.29
EP2 avg
ft3/s
.040
.056
0.045
0.033
0.039
0.027
0.039
0.033
0.039
0.037
0.033
0.026
0.034
0.030
0.028
0.030
0.022
0.030
UF3 avg
ft3/s
.022
.026
0.027
0.019
0.027
0.025
0.030
0.033
0.012
0.027
0.032
0.029
0.007
0.018
0.022
0.018
0.015
0.024
DF4 avg
ft3/s
.0003
.0005
0.001
0.001
0.002
0.000
0.000
0.006
0.001
0.000
0.001
0.002
0.008
0.001
0.006
0.009
0.003
0.002
DF5 start
s
354
1606
1624
1313
276
930.5
652
681
766
683.5
611
635
1091
3362
1018
526
1278
2079
DF5end
s
5610
4059
4109
2729.5
4372.5
3011.5
1037
1113
2500.5
3708.5
886.5
1457.5
6974
3919.5
4894
5422.5
4299.5
2459
Ratio6
UF/EP
.54
.46
0.61
0.57
0.69
0.94
0.76
0.99
0.30
0.72
0.95
1.09
0.21
0.61
0.78
0.61
0.71
0.81
There was more runoff than rainfall measured during this simulation
2 Average runoff rate from source area (EP), ft3/s
3 Average runoff rate along the toe upslope of the fence (UF), ft3/s
4 Average flow rate through the fence (DF), ft3/s
5 Start and end times of flow through the fence
6 Ratio of volume discharged at end of toe trench (UF) to volume discharged from source area (EP)
25
-------
After the flow data were collected and reduced, it was observed that discharge from the
downstream end of the toe trench was almost always significantly less than the runoff rate from
the source area. Once this was discovered, the system was examined carefully during the next
simulation, but there were no apparent locations where runoff that had been sampled could escape
prior to reaching the fence. In fact, it was considered more likely that extra water not factored
into the runoff rate could reach the fence. This was observed to happen during the very first
simulation when there was a strong north wind which blew an observable amount of water
directly onto the sheet metal. Fortunately, north winds are rare in Oklahoma and that was the
only simulation with that problem.
Even given the bulging of the fence and occasional blockage of the downstream collection trough
and lack of reliable data about flow rates through the fence, it was clear from visual observation
that the flow through the fence was not sufficient to account for the differences. Therefore, it was
concluded that the difference was due to seepage into the walls and bottom of the toe trench.
Results Based on Sediment Analysis
The first observation that should be noted regarding the sediment data is that there is a very
strong natural randomness to the processes that control sediment detachment and movement. For
example, since the slope and area of the runoff plot were constant over all the simulations and the
rate of rainfall only deviated slightly, very similar rates of sediment production would be
expected over all tests involving a particular soil. Such was not the case, even once rates of
sediment production were normalized into tons/area/unit of time/inches of rain.
However, the sediment data did reveal some trends which both supported and challenged the
existing theory and understanding of sediment movement. Table 5-2 summarizes the average or
representative sediment results for each simulation at the source area and Table 5-3 gives the
sediment results for the toe trench and through the fence.
When normalized to account for differences in length of simulation, sediment production from
the source area was greatest for sandy loam, then loam, with silty clay having the lowest. All
other things being equal, it was expected that sediment production would follow Wischmeier's
soil erodibility value, K, but sandy loam had the lowest K-value and silty clay had the highest.
This apparent anomaly was attributed to the fact that the soil was placed onto the source area
from stockpiles, making it even more disturbed than the soil on a field construction site. In
addition, the non-cohesive sandy loam soil would experience a more thorough destruction of the
soil matrix in this process than would the cohesive clays.
The concentration of sediment in the toe trench discharge also followed the same order, with
sandy loam having the highest, followed by loam, then silty clay. This is expected since the rill
erodibility factor follows that order, and more detachment of sandy loam would be expected.
However, the sandy loam also has a much higher settling velocity so there should be more
deposition. The size of the data set did not permit a statistically-based evaluation of how these
offsetting factors influenced sediment flow in the trench. The fact that the trench was filled with
stockpiled soil would make a transport capacity assessment based on theory developed for natural
streams invalid. So this issue had to be left unresolved for now.
One observation of the data that clearly followed established theory was that the concentration of
sediment in the toe trench discharge increased as slope increased. This trend occurred for all
three soils.
26
-------
Table 5-2. Sediment results for source area
Simulation date
12/18/2003
1/15/2004
3/2/2004
3/10/2004
3/19/2004
3/24/2004
4/20/2004
4/22/2004
4/28/2004
5/4/2004
5/7/2004
5/11/2004
5/18/2004
5/21/2004
5/25/2004
5/27/2004
6/2/2004
6/4/2004
Soil type
Loam
Loam
Loam
Loam
Loam
Loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Silty clay
Silty clay
Silty clay
Silty clay
Silty clay
Silty clay
Total
discharge
Ib
107
284
433
187
305
151
107
34
425
256
159
156
198
130
220
61
134
60
Average
concentration
mg/L
14899
19171
40461
19244
28540
30759
26871
9212
66110
24302
30162
36568
14621
12549
27900
6395
21066
16414
Peak
concentration
mg/L
42499
46509
122929
29000
47203
39017
42536
15284
98682
55576
48786
190540
55255
23778
49829
11035
36702
39600
Average sed
flow rates
Ib/s
0.060
0.048
0.078
0.041
0.068
0.052
0.058
0.019
0.143
0.059
0.057
0.054
0.036
0.023
0.053
0.011
0.027
0.021
Occurrence of failure through erosion of the toe trench also occurred as expected. The sandy
loam soil, clearly the most erodible, was always scoured completely away at one or more points.
For the loam, complete scour only occurred with the steep slope, and complete scour never
occurred with the very still silty clay. An additional observation was that, in general, failure
occurred if the net erosion was at least 25% of the original volume of the toe trench.
The average concentration passing through the fence followed the same order given previously,
with sandy loam being highest and silty clay being lowest. This is hard to assess, since the main
impact of the fence is through detention and settling. Much less settling of the clays was
expected, but the sandy loam started out at a higher concentration. The fence did cause a
reduction in concentration for all soil types. The greatest magnitude in terms of reduced mg/L
was for sandy loam, but the highest percent reduction occurred for loam. Sandy loam also had
the highest ratio of concentration through the fence to concentration along the fence.
The fabric type was also a factor in the concentration passing through the fence. Overall, the
concentrations passing Nilex 2127 were about 50% higher. This was expected since 2127 had an
observably looser weave and a higher coefficient of discharge, giving a lower detention time and
less settling. The fabric type did not appear to be a factor for silty clay. Since it was extremely
fine, almost no settling would occur, and the fence would have no "filtering" affect whatsoever
on the particles in suspension.
27
-------
Table 5-3. Sediment results for toe trench and passing through the fence
Simulation Date
12/18/2003
1/15/2004
3/2/2004
3/10/2004
3/19/2004
3/24/2004
4/20/2004
4/22/2004
4/28/2004
5/4/2004
5/7/2004
5/11/2004
5/18/2004
5/21/2004
5/25/2004
5/27/2004
6/2/2004
6/4/2004
Soil type
Loam
Loam
Loam
Loam
Loam
Loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Sandy loam
Silty clay
Silty clay
Silty clay
Silty clay
Silty clay
Silty clay
Toe trench
Total mass
discharge
Ib
61
104
446
218
594
511
531
969
122
204
591
426
23
112
262
52
43
44
Average
concentration
mg/L
8564
15198
75260
42764
116298
122042
195635
174206
83722
27693
120681
95282
8495
19593
47863
9052
15178
12114
Peak
concentration
mg/L
40065
22779
203479
113386
259380
246513
396115
311127
106455
101908
279870
176253
26506
51358
71155
14367
20882
25626
Average sed
flow rates
Ib/s
0.011
0.026
0.108
0.044
0.146
0.172
0.389
0.167
0.070
0.035
0.216
0.184
0.003
0.022
0.068
0.011
0.015
0.012
Net erosion
Ib
0.0
0.0
12.6
31.5
289. 11
359.61
424. 81
935.41
O.O1
O.O1
432.21
270.21
0.0
0.0
42.8
0.0
0.0
0.0
Net
deposition
Ib
45.3
179.7
0.0
0.0
0.0
0.0
0.0
0.0
303.2
52.1
0.0
0.0
175.2
18.4
0.0
9.6
91.7
16.2
Through fence
Total
discharge
Ib
0.9
0.7
0.4
3.8
52.5
3.2
2.1
34.5
0.7
0.3
6.6
22.8
20.2
1.0
19.5
25.3
4.4
0.4
Average
concentration
mg/L
4838
6505
13102
31639
103201
63218
79305
181813
3728
3860
93409
134009
6651
10805
15943
8744
6819
3386
Peak
concentration
mg/L
12744
13648
19262
44548
140986
74437
96544
197841
7053
10768
122045
158815
11971
11443
47101
16673
8133
3524
1 Toe trench failed
28
-------
Analysis of Variance and Trends
To the extent that trends in the results responded to changes in quantifiable input parameters, i.e.,
soil properties, site geometric properties such as slope along the toe, or fabric properties, analysis
of variance (ANOVA) was performed to determine if the trends were statistically significant.
The data analysis add-in to the Excel spreadsheet was used for this purpose, with an alpha level of
5%. The ANOVA results are in Table 5-4.
The trends - statistically significant or not - were also evaluated as to whether or not they
followed expectations. This assessment is shown in Table 5-5. For Table 5-5, three generic
inputs were defined - fabric type, slope along toe, and soil type. The results shown in the table
are averaged over all simulations with that generic input. With slope, trends were assessed
simply as to whether they increased or decreased as slope increased. With soil type, trends were
assessed for correlation with the most relevant soil property: d50 (mm), Wischmeier's soil
erodibility K, or rill erodibility factor (s/m). For reference, these properties are also included in
the table.
This analysis had two purposes. One was to either bolster support for established assumptions
about hydrology, sedimentation, and silt fence performance or to point out assumptions that may
require further investigation. The other was to determine qualitative trends that could be used as
part of the assessment of the design aid.
Generally, the trends observed in the field data followed what was expected. Specific comments
on each individual item are given in the table.
29
-------
Table 5-4. Analysis of Variance (ANOVA) results
Result
Ratio of runoff to rainfall
Average runoff rate from source
area, in./h
Average sediment discharge
rate from source area,
ton/acre/h
Average sediment discharge
rate from source area,
ton/acre/h per in. of rain
Average concentration at source
area, mg/L
Discharge in toe trench, ft3/s
Discharge in toe trench, ft3/s
Ratio of discharge from source
to discharge at toe
Ratio of discharge from source
to discharge at toe
Sediment discharge in toe
trench, Ib/s
Sediment discharge in toe
trench, Ib/s
Average concentration in toe
trench, mg/L
Average concentration in toe
trench, mg/L
Concentration passing through
fence, mg/L
Concentration passing through
fence, mg/L
Concentration passing through
fence, mg/L
Ratio of concentration through
the fence to concentration along
the fence
Ratio of concentration through
the fence to along the fence
Variable
Soil type
Soil type
Soil type
Soil type
Soil type
Soil type
Slope
Soil type
Slope
Soil type
Slope
Soil type
Slope
Soil type
Slope
Fabric
Soil type
Slope
F statistic
6.31
3.79
2.57
4.78
2.22
4.14
0.76
1.19
1.22
5.32
3.06
6.64
0.43
3.83
3.32
0.72
0.09
3.24
P value
0.01
0.047
0.11
0.02
0.14
0.04
0.49
0.33
0.32
0.02
0.08
0.01
0.66
0.05
0.06
0.41
0.91
0.07
F critical
3.89
3.68
3.68
3.68
3.68
3.68
3.68
3.68
3.68
3.68
3.68
3.68
3.68
3.68
3.68
4.49
3.68
3.68
Conclusion
Significant
Significant
Not significant
Significant
Not significant
Significant
Not significant
Not significant
Not significant
Significant
Not significant
Significant
Not significant
Significant
Not significant
Not significant
Not significant
Not significant
30
-------
Table 5-5. Summary of trends assessment
Result by fabric type
Concentration passing through fence, mg/L
Result by slope
Discharge in toe trench, ft3/s
Ratio of discharge from source to
discharge at toe
Sediment discharge in toe trench, Ib/s
Average concentration in toe trench, mg/L
Concentration passing through fence, mg/L
Ratio of concentration through the fence to
concentration along the fence
Result by soil type
Ratio of runoff to rainfall
Estimated Curve Number
Average runoff rate from source area, in./h
Wischmeier's K values
650 of soil, mm
Result by soil type
Average sediment discharge rate from
source area, ton/acre/h
2127
53679
Low
0.021
0.592
0.028
80266
4856
0.318
Loam
0.83
97.7
2.33
0.32
0.0555
Loam
5.70
2130
31985
Moderate
0.022
0.673
0.096
70282
48269
0.738
Sandy loam
0.74
94.1
1.87
0.14
0.1482
Sandy loam
5.66
Steep
0.026
0.794
0.159
47726
75371
0.692
Silty clay
0.62
88.2
1.56
0.31
0.0051
Silty clay
2.65
Reasonable - 2130 has a tighter
weave and a lower coefficient of
discharge.
All other things being equal,
discharge in a channel increases
as slope increases. Here,
however, discharge is also
controlled by the amount of
discharge from the source and
the amount of seepage. Since
there should be a lower travel
time with a higher slope, and
therefore less opportunity for
seepage, this result is
reasonable.
Reasonable - the higher slope
would give a shorter travel time
and less opportunity for seepage.
The higher slopes lead to higher
velocities with more potential for
scour.
This result does not agree with
what would be expected.
Not sure, slope might be a factor
in this.
Not sure, slope might be a factor
in this.
This agrees with visual
observation of the soil structure.
The loam formed a very tight
surface, the sandy loam was
more granular, and the silty clay
was very clumpy with numerous
macro pores that had to be filled
before runoff occurred.
Reasonable for same reason as
above.
Reasonable for same reason as
above.
Measured of computed property -
included for reference.
Measured of computed property -
included for reference.
Observation of the soils indicated
that the sandy loam was the most
easily erodible, loam was
moderately erodible, and silty clay
was erosion resistant. This result
does not follow the Wischmeier's
31
-------
Average sediment discharge rate from
source area, ton/acre/h per in. of rain
Average concentration at source area,
mg/L
Discharge in toe trench, ft3/s
Ratio of discharge from source to
discharge at toe
Rill erodibility factor, Kr, s/m
Sediment discharge in toe trench, Ib/s
Average concentration in toe trench, mg/L
Concentration passing through fence, mg/L
Result by soil type
Ratio of concentration through the fence to
concentration along the fence
2.44
25512
0.024
0.64
0.00283
0.085
18716
8725
Loam
0.56
4.12
32204
0.027
0.80
0.00361
0.177
116203
82688
Sandy loam
0.64
1.55
16491
0.017
0.62
0.00111
0.022
63354
37084
Silty clay
0.55
K values which were computed
from the soil texture.
Result agrees with visual
observation of the soil, but does
not follow the K-value. This result
does correlate with the 650 of the
soil.
Result agrees with visual
observation of the soil, but does
not follow the K-value. This result
does correlate with the 650 of the
soil.
Soil type is a factor in this in grain
roughness and in losses due to
seepage. This result does not
follow what we would expect with
grain roughness, since the
coarser grained sandy loam
would provide more resistance to
flow. However, the sandy loam
also appeared to have a lower
oss due to seepage since its non-
cohesive structure provided fewer
macro pores for significant
seepage.
Reasonable because the sandy
oam appeared to have less loss
due to seepage.
Measured of computed property -
included for reference.
Reasonable, follows rill erodibility.
Reasonable, follows rill erodibility.
To the limited extent that the
fabric acts as a strainer, we would
expect less of the coarser
material to pass through, which
did not occur. Plugging of the
openings in the fabric is another
way that the soil can affect the
flow through. Here, we would
expect the coarser material to
plug the openings more than the
clays, which would pass through
easily. This supports
observations made in the flume
testing that plugging is not a
significant factor when the flow is
parallel to the fence.
32
-------
Chapter 6
Modeling Silt Fence Performance
Background
A key element of this project was development of a mathematical model of silt fence
performance. The purpose of the model development was to provide a means of predicting silt
fence performance under a wider range of site and hydrologic/hydraulic conditions that could be
simulated during the field laboratory tests. The mathematical model also forms the basis of the
design aid.
Ideally, a model of silt fence performance will contain the following components. However,
limitations in available data for validation and a need for simplicity for the user community made
some of these elements impractical.
Hydrology. Predict runoff volume and rate, sediment yield, and sediment size distribution of
flow arriving at the silt fence as a function of soil type and cover characteristics of the source
area, along with rainfall characteristics.
Hydraulics. Predict flow through the fence and along the fence as a function of the type of
fabric, incoming flow rate, and slope along and toward the toe of the fence.
Sediment transport. Predict sediment transport along the toe and through the fence; predict
lateral flow erosion along the fence as a function of flow characteristics, soil properties, and
slope.
Sediment trapping. Predict the total sediment loading to the fence, total weight that is
discharged at the downstream end of the toe of the fence, and total weight passing through the
fence. If a simple impoundment is created by extending a length of silt fence up the slope at
the downstream end, predict the trapping efficiency of the impoundment.
As a first step in model development, the functionality of several existing water and sediment
routing models was reviewed to determine if there was one that might be easily adapted for this
project. Models considered included WEPP (Lindley et al., 1998), SEDIMOTII (Wilson et al.,
1984), and SEDIMOT III (Barfield et al., 1996). SEDIMOT III was selected because in addition
to the required hydrology and overland erosion components, it also included a concentrated flow
channel erosion component and a component that routes a hydrograph and sedigraph through an
impoundment and generates effluent concentrations along with trapping efficiency.
To create a preliminary model, some minor modifications to SEDIMOT III were made, primarily
modifying the channel flow component to account for flow through the fence (in order to use the
channel erosion routine to simulate the flow along the toe of a silt fence) and modifying the
output to print out the erosion calculations for each time step. Data from the first three
construction site visits was used as input data to see if this preliminary model performed in a
reasonable manner, and concluded that it did an adequate job of predicting runoff and sediment
33
-------
yield from the contributing drainage area. However, the predicted channel erosion did not agree
with what was observed in the field.
From this, it was concluded that the channel erosion routines based on equilibrium channel
geometry and excess shear concepts (Storm, 1991) and developed for natural streams were not
applicable. This was considered due to the differences between natural soil and the excavated/re-
emplaced soil in a silt fence toe trench. The preliminary model was used to design some of the
details of the field laboratory test plot, specifically, to size the source area samplers and
conveyance system and as an aid in planning sample collection intervals.
Since there was no off-the-shelf model to use, the mathematical model was programmed to
operate within an Excel spreadsheet in order to make it adaptable to a simple design aid format.
This also simplified coding and de-bugging.
This document is intended for the user community, therefore the inputs and outputs for the design
aid described in Chapter 7 are those normally used by engineers and other practitioners in the
erosion control profession. This is also followed in the model development described in this
chapter.
Development of Model Components
In general, the model was constructed as a quasi steady-state model. While it does do
computations in time steps, the inputs to each time step are identical, so the model converges to a
steady state solution. For example, a uniform intensity of rainfall is assumed over the duration,
so the depth of rainfall is equal for all time steps. The steady state approach was considered
reasonable because an equilibrium or steady state condition to develop very quickly in most of
the field laboratory tests was observed.
Hydrology Component
Elements of the hydrology component were adopted almost entirely from Sedimot III, and are
also included in many other widely-used hydrology models. The following paragraphs describe
the hydrology component.
Runoff Volume
Runoff volume as a function of rainfall and land cover is computed using the NRCS curve
number equation (NRCS, 1969). The cumulative rainfall in each time step is the precipitation
input. Since steady-state conditions were almost always observed to happen within 1 h during the
field testing, a storm duration of 3 h with 0.1 h time steps was selected. This is a storm of
sufficient length to create equilibrium in the flow through silt fence, as discussed later.
Incremental rainfall in each time step is obtained by dividing the 3-h storm depth by 30.
Cumulative depth is then obtained by summing.
The NRCS has developed standard precipitation distributions which are commonly used for
hydrologic investigations. These storms require a 24-h return period precipitation and then can
generate a distribution for any duration less than 24 h with the same return period. To simplify
user inputs, the model works from a standard 24-h return period storm as a user input. The model
then determines the 3 h storm with the same return period and uses that for runoff and peak
discharge calculations.
34
-------
Cumulative runoff, Qp (in.) in each time step is determined from the cumulative rainfall, P (in.),
and curve number (CN), which is a parameter that reflects soil type and land use, as:
Where, S (in.) is potential abstraction from rainfall which is defined by the curve number as
CN
The quantity 0.2S is commonly used as the initial abstraction before start of runoff, so QP is equal
to zero for values ofP where P < Q.2S. The value of QP represents the inches of rainfall that are
converted to runoff from the source area. Total volume of runoff is obtained by converting the
inches to feet and multiplying by the land area. Computing with area in acres gives runoff
volume in acre-feet and computing with area in square feet gives runoff volume in cubic feet.
Both are determined for use in different components of the model. The incremental volume in
cubic feet is determined for each time step by subtraction.
Since the land areas are small, runoff can be assumed to be instantaneous, as opposed to being
routed overland using a unit hydrograph or kinematic model. Therefore, the runoff rate from the
source area, qp (ft3/s), at any point in time is the rate of change of runoff volume, or:
dQP
and the maximum runoff rate will be
\~rin 1
(6.4)
Sediment Yield
Sediment yield from overland erosion is computed with the modified universal soil loss equation
(MUSLE) (Williams and Brendt, 1972), or:
Y = 95 Qpqp:max ° 56 K[LS][CP] (6.5)
where K is an empirical soil erodibility, LS is the dimensionless length slope factor which is
determined from slope and slope length (Haan et al, 1994) and CP is a dimensionless cover and
practice factor that accounts for the impact of cover and compaction on sediment yield. This
equation uses the computed values of peak runoff rate and runoff volume, the length and slope
toward the fence, and a soil loss parameter, Wischmeier's K (Wischmeier et al., 1971), that is
input by the user. Recently published NRCS county soil surveys have tabulated values of K.
Otherwise, K can be determined from soil texture by using the Wischmeier et al. (1971)
nomograph (Figure 6-1).
35
-------
UR£: With opproprlate doto,
enter «col« at left with % Silt + vfc (.002 - 0.1mm) ond
proceed to point* repre»«nting th* »o8*» K tond (O-iO-2-Omm),
% organic motter, structure, ond permeability, in thot sequence.
fnterpafate between plotted curve*. T>ie dotted line Hu*trat*s
procedur* for o toff havtng: ei+vfe 65X sand 5%, OM 2-6X,
itructure 2. permeabHtty 4, Solution: K-0.3V ton tflHtf........
or ,04 ,, t!qfl*hE . niHidr«ds oc-n-tonf*
Figure 6-1. Wischmeier et al. (1971) nomograph for determining K factor
The final MUSLE parameter, a cover factor, is also input by the user. The recommended range
for "bare soil, undisturbed except scraped" is 0.66 to 1.3 (Transportation Research Board, 1980).
If the site cover is soil that has been recently placed from a stockpile, use of Figure 6-2 to
determine cover factor is recommended (Haan et al., 1994). This accounts for the fact that, over
time, soil that has been placed from a stockpile will become less erodible.
For calculation of sediment transport, an average particle diameter, d50 (mm) is needed to
calculate sediment transport. Since the drainage areas contributing to a silt fence are typically
small, soil was described by a single representative diameter, d50 (mm), determined by the eroded
size distribution. Also, to predict trapping in the impoundment at the end of the silt fence,
particle size classes are needed. This can be based on actual measurements of eroded size
distribution (Haan et al., 1994, Chapter 7) or estimated by the CREAMS model that is widely
used to estimate eroded size distribution. Although the CREAMS equations are not as accurate as
might be desired for eroded size distribution from construction activities (Barfield et al., 1988) it
is the only predictive technology currently available. A study of eroded size distribution from
construction sites is needed to fill this information gap. Developing such a model was beyond the
scope of this project and not possible with the relatively small number of field trials conducted,
therefore the CREAMS model will be used with a user option to input a measured eroded size
distribution.
36
-------
O
Q
05
CO
2.6
2A
2.2
2.0
1.8
1.6
1,4
1.2
1.0
08
f
V
?
o
V
\
v
\
S
\
\
>
i,
\
O TopsoO
x Subsoil
S
W'
"m;
^
^
-*4
n
0
6
12
Time After Reconstruction
With Stockpiled Soil (Months)
Figure 6-2. Cover factor for use of stockpiled soil
Hydraulic Component
Flow along the toe is computed based on Manning's equation (Haan et al., 1994). A triangular
channel geometry, wherein the vertical silt fence is one leg of the triangle and the overland slope
is the other leg (see Figure 6-3) is used to compute area and hydraulic radius. A typical bare soil
value of 0.025 is used for roughness factor (Manning's «).
To do the routing calculations along the silt fence toe, it is convenient to have a simple
relationship giving discharge along the toe as a function of head. Using Manning's equation and
the triangular geometry shown in Figure 6.3 the flow rate along the toe of the fence, q (ft3/s), can
be expressed as:
q = KqHSA (6.6)
Where, Kq is a constant for a given combination of Manning's n, slope along the toe, and slope
leading to the fence and H is the depth at the fence as shown in Figure 6.3. To compute Kq, the
model includes a subroutine which finds q for values of H ranging from 0 to 1 ft and then uses
linear regression to find the slope of the line formed by the q vs. H813 data points.
37
-------
Fence
Direction of
overland flow
H
^^ \
Cross section
area of flow long
toe
Figure 6-3. Triangular channel geometry
Flow routing along the toe is based on a mass balance wherein the basic continuity relationship is:
Volin-Volout=ASt (6.7)
Where, Volin and Volout (ft3) are inflow and outflow volumes in a time step and ASt is the change
in storage along the toe of the fence in the time step. Volinis the sum of the volume exiting the
upstream section plus the overland runoff. Essentially, a four-point finite difference grid is used
for the computations, as shown in Figure 6-4. The current version of the model is implemented
using an Excel spreadsheet. To compute flow along the toe, the flow length along the toe is
divided into two reaches, giving three nodes along the fence. The model solves for the head at
each node in each time step. The boundary conditions are H=Q for all x at t = 0 and H= 0 for all
38
-------
T
At
o
(0,1)
o
(1,1)
(0,0)
(1,0)
Ax
Figure 6-4. Schematic for four point grid solution matrix
The model is actually an explicit system of nonlinear equations, i.e., the solution for any property
such as Voljn, Volout, or ASt at Node (1,1) (see Figure 6.4) can be explicitly calculated from known
values at Nodes (0,0), (0,1), and (1,0), all defined as boundary conditions. Once H(u) and the
other values at (1,1) are computed, then the values at (1,2) and (2,1) are functions of known
values. It is therefore possible to go through the grid sequentially and solve for each cell. Using
cell (1,1) as an example, the nonlinear equations solved are:
H
/ At + qof,At
(6.8)
Where, qof,i is the overland runoff rate for one-half of the total area during time step 1. The
computation for Vol0ut includes the discharge exiting at the downstream node, flow through the
fence and losses from seepage:
Vol
0ut(V)
H
f 1j0
At + qF At
(6.9)
Where, Kq was defined earlier as the constant relating flow velocity, channel roughness, and
channel geometry for flow along the fence, Vseep (ft3) is the volume of flow that seeps under the
fence, and qF (ft3/s) is the flow rate through the fence, or:
qF = KFCS (HO/ + H0/ + H,/ + H,/) Ax
(6.10)
Where, KF is the fence constant which relates flow through the fence to the depth of impounded
water for the static condition and Cs is a dimensionless parameter which accounts for the impact
of parallel flow along the fence on the flow through the fence, or:
39
-------
(6.11)
Where, F (ft/sec) is the velocity in the toe trench, Sc is the slope of the toe trench, and Kc, PI and
P2 are empirical coefficients. Values for Cs, Kc, KF, PI and P2 are all parameters that are
functions of the fence material and require calibration with laboratory or field data. For Nilex
2130 or 2127, the following values have been determined based on flume studies:
^, = 0.0659 for Nilex 2127; 0.0306 for Nilex 2130.
Kc = 1.906
Pl = 0.6171
P2 = 0.6224
)L SCCptlgC 1USS, V seep ^1*
computed as:
The volume of seepage loss, Vseep (ft), is a function of the overland runoff volume and is
(6.12)
Where, KA is the adjustment factor for the ratio of volume discharged along the toe to overland
runoff volume, VOL (ft3)- This parameter was estimated using the field data as 0.6863. At this
point, the best estimate is the average of all the adjustment factors that were backed out of the
computation using the field data. This will be a user input parameter for the model.
From geometry, the change in storage is computed as:
AS, = 0.5| ^-(Hai;2 -Hao;2) + -l-(Hf0,i;2 -Hf0,o;2) |Ax (6.13)
Where, SL is the slope perpendicular to the fence. The model uses a solver in the spreadsheet to
find the head at each node that minimizes the squared error in the continuity equation written as:
Volln-Vol0ut-ASt=0 (6.14)
Equations (6.6) through (6.13) are used to define Volin, Volout and ASt. Once H (U) is known,
based on the solver routine, values of g/(i,i) and #(i,i), are determined and a new routing time step
generated.
Sediment Component
In order to solve the equations that define sediment load into the impoundment at the end of the
silt fence, it is necessary to define the inter-relationships between incoming sediment load,
detachment potential, and deposition. Typically, this is done with a process based sediment
transport equation combined with a detachment/deposition routine. This was tried initially in this
project, however, as noted earlier, relationships based on traditional channel erosion and transport
theory did not perform well in making predictions of sediment transport along the toe of the silt
fence. Therefore it was necessary to develop these relationships using the field data. Based on an
analysis of the data collected, the parameter least influenced by the scale of the tests was the
average concentration of sediment in the toe trench discharge.
40
-------
Concentration of sediment in the toe trench
Both an Analysis of Variance and visual observations noted during the field trials indicated that
the average concentration of sediment in the toe trench discharge was a function of slope and dso
(mm), so a relationship of the form shown in equation (6.15) was estimated using the field data,
or:
= 3.254(1 OOSJ2440n OOOcf5J1
(6.15)
Where, CTavg (mg/L) is the average concentration in flow along the toe and Sc and d50 are as
previously defined. With more data it should be possible to develop a more processed-based
model using channel erosion and transport capacity theory with adjustments for the fact that the
toe trench bed was backfilled material, as opposed to being natural stream bed. However, there
just was not enough data to attempt that at this point. Figures 6-5 and 6-6 show the predicted vs.
observed values and a comparison of individual simulations. The value for dso can come from a
measured eroded size distribution or can be estimated from the CREAMS relationships (Knisel,
1980).
250000
200000
O)
1 50000
a>
1 00000
50000
0 20000 40000 60000 80000 100000 120000 140000 160000 180000
Predicted, mg/L
Figure 6-5. Comparison of observed and predicted concentrations in toe trench using equation (6.15)
41
-------
250000
4 6 7 8 9 10 11 12 13 14 15 16 17 18
Simulation Number
Figure 6-6. Comparison of observed and predicted concentrations by simulation number (given in
Table 4-2)
Equation (6.15) actually is intended as a method to predict transport capacity. If it is assumed
that the flow exiting the toe strip is operating at the sediment transport capacity, the error would
not be great, based on our observations. Based on this assumption, the deposition or scour in the
toe trench can be estimated from the difference in source area sediment yield and flow from the
toe trench into the impoundment.
Sediment discharge in the toe trench
Using equation (6.15) for sediment yield, sediment discharge rate in the toe trench is then
computed by multiplying the concentration expressed as lb/ft3 by the toe trench discharge in ft3/s.
The total mass of sediment transported along the toe to the outlet, Wst (lb) during a time step is
computed as:
(6.16)
Where, qstl and qst2 (Ib/s) are the sediment discharge rates in sequential time steps.
Sediment discharge through the silt fence
Flow through the fence, qFis obtained from the hydraulic routing. To convert this to a sediment
discharge, the flow rate must be multiplied by sediment concentration. In general, one would
expect the concentration flowing through the fence to be approximately equal to the suspended
load concentration. Unfortunately, there are no well documented suspended load equations for
soil erosion. In the absence of such a relationship, a user input is put into the model to allow user
selection for the parameter. For the field studies conducted, the average concentration through
the fence based on the field data was found to be a fraction of the concentration in the toe trench.
42
-------
This fraction (Fc) was a function of the fabric type, with the fractions for Nilex 2130 and 2127
being 0.415 and 0.749, respectively. As with the toe trench concentration, it is hoped that if more
field data becomes available, then a more process-based model can be developed to define this
parameter. The weight of sediment discharged through the fence is then computed in the same
manner as for the toe trench, or:
-------
Where, TET (fraction) is the total trapping efficiency measured over the runoff event, MSP andMST
(Ib) refer to the mass of sediment discharged from the source area and mass of sediment moving
from the toe trench to the impoundment. The assumption can be made, as shown by Hayes et al.
(1984) that the deposited material will be the largest particles, therefore the size distribution must
be adjusted accordingly. The new fraction finer for each size class, PFItt, at the exit to the
impoundment is given by:
PF/1=/Wax[(PFP1-TE7),0.o]
PF/2 = Max{PFP2-Min[(PFP, -TE7),0.o],0.o}
PF/3 = Max(pFp3 -Max[PFp2 -Min[(PFp1 -TE7),0.o],0.o})
PF/4 = Max(pFp, -Max(pFp3 -Max{PFp2 -M;n[(PFp1 -TE7),0.o],0.0
PF,K =Max
PFP5 -Max/PFp4 -Max(pFp3 -Max{PFp2 -M;n[(PFp1 -TE7),0.o],0.0
(6.19)
Impoundment Component
The geometry of the impoundment included in the model represents what is created by extending
a length of silt fence upslope at the downstream end of the toe. The area relationships are based
on the assumption of a uniform slope along the toe and a uniform overland slope. Therefore, the
impoundment bottom can be modeled as a plane with the corner being the lowest point. Figure 6-
7 is a schematic showing that H now refers to the maximum depth of water at the lowest point in
the impoundment.
To develop the impoundment computations required development of depth vs. area and depth vs.
outflow relationships, along with a trapping efficiency function. The equation for surface area as
a function of the maximum impoundment depth, HIM (ft) in the lower corner is:
tan 9
2Sc(-Sc+SLtand)
(6.20)
Where, 6>is the angle between a line along the bottom of the silt fence and the extension, Sc, is the
slope along the toe of the silt fence, and SL is the slope perpendicular to the silt as shown in the
schematic in Figure 6-7. The volume is then found by integrating A(h)dh from h = 0 to h = HIM,
or
V(H } =
tan 9
(6.21)
To account for the fact that the head on the silt fence varies from a maximum oftfm in the corner
to zero where the water surface intersects the ground (see Figure 6-8), the equation for flow
through the fence was integrated as:
QFT = JKF [/7(x)J dx = KF
dx =
= f KFHlf
(6.22)
44
-------
TOP VIEW
Extension
Impounded
surface area
Fence and Posts
SL - overland
slope
perpendicular
,. to fence
\ ,r orientation
(0,0,0)
S - slope along toe I
H - depth
in corner
Figure 6-7. Impoundment geometry
Where, QPT is the total flow through the fence panel (ft3/s, not ft3/s/ft), KP is the discharge
coefficient for the fabric type, and HIM and L are as shown in the figure. For the main silt fence, L
is equal to HIMISC. For the extension, L is equal to
1 + tan2 6
Sc + SLtan0)2
The final equation for the impounded flow through the silt fence is therefore
1
1 + tan2 9
(-Sc+SLtan0)z
(6.23)
(6.24)
45
-------
Note that for the extension to actually create an impoundment, tan9l(-Sc + SL tan6>) has to be
greater than zero.
x=0
x=L
dx
Figure 6-8. Schematic of flow through silt fence around impoundment
After running a series of full impoundment simulations, it was observed that the depth in the
impoundment reached steady state, i.e. inflow equal to outflow, within a very short time, always
in less than 10 min. The impoundment hydraulics was therefore modeled with steady-state
conditions, as follows:
1. The peak discharge at the downstream node of the toe trench was the constant inflow rate.
2. The depth (H1M) was computed using equation (6.24), surface area and volume were then
computed based on HIM.
3. The overflow rate velocity, Vc, was then computed as peak discharge divided by surface area
at HIM-
Item 3 introduces the concept of an overflow rate, a parameter that is used in calculations of
trapping efficiency. The overflow rate is a settling velocity that will just allow the particle to
settle from the top to the bottom of a rectangular basin with steady quiescent flow. The concept
has been extended to non-rectangular channels and referred to as dynamic removal efficiency
(Driscoll et al., 1986). The Driscoll et al. model was used to compute the trapping efficiency.
This model requires a parameter, (3, which is a performance factor, as follows:
(3=1, very poor performance
(3 = 2, average performance
(3 = 3, good performance
46
-------
(3 >5, very good performance
The user inputs |3 and the model computes trapping efficiency for each particle class, TEt as:
7E,= 1-
1 V
-^-| |x100%
P Vc
(6.25)
Where, Vs:i is the settling velocity for particle class /'. The total trapping efficiency over all
particle size classes is:
TETOT=j^(TE)l(PF)l (6.26)
/=i
Finally, the model also warns the user if the water surface is above the end of the extension, the
water discharges around the silt fence.
Validation of Model
Model validation included both a qualitative and quantitative component. For the qualitative
assessment, the model responses to changes in input values were compared to the trends reported
in Chapter 5. To complete the quantitative component, the conditions present for the field testing
were inputted into the model and model outputs were compared to the observed results.
Qualitative Assessment
The model output was averaged according to the generic inputs described in Chapter 5, and the
averages were compared to the actual field averages to see if the trends were preserved. Table
6.1 gives the results for selected comparable outputs. The outputs selected for comparison were
chosen because in the model they are functions of those generic input values.
Table 6-1. Assessment of trends in model output
Result by fabric type
Concentration passing through fence (mg/L)
Field Lab - increased as discharge coefficient increased
Model - increased as discharge coefficient increased
Result by slope
Discharge in toe trench, (cfs)
Average concentration in toe trench (mg/L)
Concentration passing through fence (mg/L)
Field Lab - increased as slope increased
Model - increased as slope increased
Field Lab - decreased as slope increased
Model - increased as slope increased
Field Lab - increased as slope increased
Model - increased as slope increased
Result by soil type
Average concentration in toe trench (mg/L)
Concentration passing through fence (mg/L)
Field Lab - ascending order - red clay, black clay, loam
Model - ascending order - black clay, red clay, loam
Field Lab - ascending order - red clay, black clay, loam
Model - ascending order - black clay, red clay, loam
The concentration through the fence as a function of fabric type preserved the trend in the field
data. This is due to the way the model was developed. The computed discharge in the toe trench
as a function of slope also followed the trend in the field data. However, the average
concentration did not follow the results seen in the field. Those field results were not as expected
(see Table 5-4), and the trend shown here does follow what we would expect to see. The model
47
-------
trend in concentration passing through the fence as a function of slope followed the field-
observed trend. The model prediction is a function of fabric type and concentration in the toe
trench.
The results as a function of soil type were not as consistent as the results by slope or fabric type.
For both concentrations along the toe and through the fence, the loam soil produced the highest
value, both by model and in the field. However, the trends for the other two soils were reversed.
The model predictions followed the trend in the d50- the field observations did not. The model
observations are also explicitly a function of slope which may explain why the trends in the
averages do not agree.
In general, the model did a reasonable job of following the trends in the field data, with three
comparisons in complete agreement and two with peaks in agreement. For the one that
completely reversed the trend, it was noted earlier that the field observed trend was not as
expected.
Quantitative Assessment
The predictive ability of the model was also assessed through a comparison of model output and
field-observed values, using the 14 simulations that had measured eroded dso values. The outputs
displayed in the design aid were selected for this comparison. Depth of water and accumulated
sediment along the toe were not compared because this information was not collected as part of
the field studies. Also, since the observed discharge through the fence was not considered
representative of what would be seen at a field site (see discussion in Section 5), no comparisons
requiring observed flow through the fence were made. Also, since it was not a part of this project
to collect impoundment data, the impoundment functions could not be checked.
In general, the hydrology functions performed well, particularly at predicting runoff rate and
volume. Figure 6-9 is a plot of observed vs. predicted runoff rate. The root mean square error
(RMSE) in runoff rate was 0.0045 ft3/s, the average observed value was 0.032 ft3/s and the
average predicted value was 0.035 ft3/s.
For sediment yield, using a cover factor of 1.0 resulted in highly under-predicted values. The
calibrated value of cover factor (over all the simulations) was 1.7, which is in line with the value
in Figure 6-2 for emplacement of stockpiled soil. The soil for the field tests was placed from
stockpiles, and the tests were run anywhere from almost immediately to about three months after
the soil was placed. Figure 6-10 shows the predicted vs. observed values. The average observed
and predicted values were 0.081 and 0.088 tons, respectively. Further calibration of the model
with the field data was not deemed advisable because as is, the model predicts in the correct order
of magnitude and responds in changes to source area slope and soil type in line with accepted
theory of overland soil erosion.
48
-------
0.06
0 0.01 0.02 0.03 0.04 0.05 0.06
Predicted, ft3/s
Figure 6-9. Model validation - observed vs. predicted average runoff rate from source (ft /s)
0.25
0.2
2 0.15
0)
0.1
0.05
0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Predicted, ton
Figure 6-10. Model validation - observed vs. predicted average sediment yield from source (ton)
49
-------
The average discharge in the toe trench was generally under-predicted, with an average observed
value of 0.022 and a predicted value of 0.015. Figure 6-11 is a plot of the data. It was previously
noted that the seepage model was based on limited data, and there is no additional data for further
calibration. The model was determined acceptable since the actual errors were small and it
responded to changes in input values the way it was expected.
0.035
0.03
0.025
'C 0.02
a
0)
> 0.015
0)
.Q
O '-''-'''
0.005
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
Predicted, ft3/s
Figure 6-11. Model validation - observed vs. predicted average discharge in toe trench (ft3/s)
The model produced reasonable results for average concentration in the toe trench. Figure 6-12
shows observed vs. predicted. The average observed value was 69,594 mg/L and the average
predicted value was 58,049 mg/L. Predicted total weight of sediment discharged downstream
was predicted fairly well, with a few outliers as shown in Figure 6-13. The average observed and
predicted values were 293 and 185 Ib, respectively.
50
-------
250000
200000
0)150000
o"
> 100000
50000
0
0 20000 40000 60000 80000 100000 120000 140000 160000
Predicted, mg/L
Figure 6-12. Model validation - observed vs. predicted average concentration in toe trench (mg/L)
1200
1000
T3
0)
to
.Q
800
600
400
200
0
0
100
200
600
300 400 500
Predicted, Ib
Figure 6-13. Model validation - observed vs. predicted sediment discharged at end of toe (Ib)
700
51
-------
Predictions of net erosion and deposition were fair. The correct type - erosion or deposition -
was predicted correctly for all 14 simulations evaluated. Figure 6-14 shows the observed vs.
predicted. Since there are zero values, comparing the averages for this output is not very
informative. The model uses the net erosion as an indicator of failure by scour at the toe of the
silt fence. In this respect, the model performed quite well. The correct phenomena - failure or no
failure - was predicted for 11 of 14 simulations. With the three that were incorrect, for two of
them failure was predicted when it did not occur. There was only one instance where failure
occurred in a fence that was predicted to stay intact. This is a good attribute for a design aid in
that predictions are mostly conservative and include a factor of safety.
1000
900
800
700
T3
a)
a>
to
.Q
o
200
100
0
4 Net Erosion
Net Deposition
Equal
0
50
100
150
350
400
450
200 250 300
Predicted, Ib
Figure 6-14. Model validation - observed vs. predicted net erosion/deposition (Ib)
There was fair agreement for concentration passing through the silt fence. Observed vs. predicted
values are in Figure 6-15. This computation is a function of several intermediate computations
and error in this output would be compounded by even small errors in intermediate computations.
Still, the overall magnitude of the results was good, with average observed and predicted
concentrations of 42,832 and 42,658 mg/L, respectively.
52
-------
200000
160000
140000
J
3)120000
-100000
<5
W 60000
.Q
o
20000
0
20000 40000 60000 80000
Predicted, mg/L
100000
120000
Figure 6-15. Model validation - observed vs. predicted concentration of sediment in flow through
fence (mg/L)
The primary objective of the model validation/calibration was to have the model produce results
with a reasonable magnitude, but not to calibrate to the point that the model responses with
respect to changes in input values did not agree with accepted theory and practice. This model
produces, on average, predictions that agree well with the field observations, and generally
responds to changes in inputs in the appropriate manner. Therefore, it is concluded that the
modeling objectives were accomplished and the model can serve as the basis for a design aid.
53
-------
Chapter 7
Design Aid Spreadsheet
Background
The design aid spreadsheet (attached herewith) evaluates the following aspects of a silt fence
installation:
Duration of a specified rate of rainfall that will result in failure due to scouring of the toe, or
indication that failure by scour will not occur within 3 h.
Average depth of sediment deposition that occurs behind the fence. Since fences are installed
at different heights, the designer can use this information to determine the number of storms
the fence can withstand before the accumulated sediment becomes a problem and clean-out is
required.
Total pounds of sediment that will be discharged at the downslope end of the fence. Ideally,
the downslope end of the fence will form an angle up the slope to create an impoundment or
the flow will be directed to a suitable location, such as a sediment trap.
Total pounds of sediment that will be discharged through the fence.
Trapping efficiency if the toe of the downslope end of the fence is angled up-slope to create
an impoundment.
To execute the design aid, the user enters the information specified in Table 7-1.
User Instructions
The user opens the tab labeled Input and Output to enter the data described in the preceding table.
All of the cells highlighted in blue need to have entries. If the fence will not be extended uphill to
create an impoundment, the default values can be left in the impoundment data section. Once the
data are entered, the user switches to the Hyd routing page to solve the flow hydraulics
component. For now, this is executed by calling up the Excel Solver using the Tools - Solver
menu. The solver is programmed to do the computations. The user should click on the Solve
button. The highlighted box with the word Result gives the squared error in the continuity based
on the assumed head. Repeating the Solver computation will decrease the error. It is
recommended that the user repeat Solver until the error is less than 0.001 or until stops doing
additional iterations.
The user can then go back to the Input and Output page to view the results. The user may then
adjust the input parameters to show the impact of management measures on the silt fence
performance. Changing the cells highlighted in green requires repeating the Solver.
54
-------
Table 7-1. Summary of user input to design aid
Parameter
Hydrology information
24-h rainfall for selected return interval
Curve number
Length up slope
Width along fence
Slope to fence
Slope along fence
Toe trench width
Toe trench depth
Soil information
Wischmeier's K
dso
Cover factor
Eroded size distribution OR
Sand-silt-clay for CREAMS
Fabric information
Fabric type - Nilex 2130 or Nilex 2127
OR discharge coefficient
Impoundment information
Angle that extension makes with toe
Length of extension
Performance factor
Units
in.
ft
ft
ft/ft
ft/ft
ft
ft
English Units
mm
Degrees
ft
Example
A residential subdivision includes 4 building lots along 400 ft of road frontage. The fronts of the
lots slope toward the road, and the drainage divide is 20 ft from the edge of the road. A silt fence
will be used to protect the road from the construction site sediment. The slope toward the road is
8% and the slope along the road (and along the fence) is 5%. Using a 3-h rainfall depth of 2.5
in./h, evaluate the suitability of using silt fence.
Additional properties required are listed in Table 7-2.
55
-------
Table 7-2. Site properties for design aid model
Property
Curve number
Toe trench width
toe trench depth
Fabric Data
Put X next to type
Fabric type- Nilex2130
Wischmeier's K
% sand
% silt
% clay
Cover factor
Angle of extension upslope
Length of extension upslope
Beta performance factor
Value
96
0.5
0.5
0.15
20
35
45
1.7
90
4
2
Units
ft
ft
English units
degrees
ft
Figure 7-1 shows the data input screen. After executing the routing, the output can be viewed on
the same page of the spreadsheet, as shown in Figure 7-1. The output shows that the fence is not
only ineffective in trapping sediment, the presence of the fence and the concentrated flow along
the toe results in a greater release of sediment than would occur with no fence.
The user should also note that there is a warning that the water elevation in the impoundment is
higher than the end of the extension, meaning more sediment lost as flow goes around the end of
the extension. This can be corrected by increasing the length until the warning disappears. In
this case, an increase from 3 to 4 ft is sufficient.
The model can be used to assess modifications to the design. For example, the fence is re-
positioned to lower the slope along the fence to 3%. This increases the overall trapping
efficiency to 70.3%. With the lower slope, there is less scour along the toe and a lower
concentration of sediment in the flow along the toe and less sediment discharged into the
impoundment. There is also less sediment discharged through the fence because the
concentrations are lower.
If the looser, Nilex 2127 fabric is used, overall trapping decreases to 51%. This is due to the
higher concentration and discharge through the fence and the higher rates of discharge from the
impoundment.
The soil in this example is a fine-grained soil, and failure due to scour of the toe trench is not
expected. For the rainfall selected, a deposition of 0.09 ft is expected. Typically, clean-out is
recommended when the height of sediment reaches one-third the height of the fence. For a
typical 12 to 18-in. high fence, after 3 to 5 similar storms, maintenance should be considered.
These examples illustrate the use of the design aid to assess the overall trapping efficiency of a
silt fence installation and use as a management measure.
56
-------
:S] File Edit View Insert Format Tools Data Hindow Help
. Arial .10 - B / O : WE. M m g $ % , too fSi f!Oi[3'5'>-.^-A.-= =
Net deposition 3ob4
Silt fence
Fabrc selected Nilex 2127
Average concentration through fenct 4421 1
Sediment through fence 1935
cfs
tons
cfs
ft
mg/L
hours
ft
Ibs
Ibs
Ibs
mg/L
Ibs
mpoundent
Trapping efficiency 9.57 percent
Surface area 6.91 square feet
Depth 0.18ft
Discharge through fence 0.02 cfs
,
hi I HT
Ready
Figure 7-1. Screen capture of spreadsheet input and output
57
-------
References
Journal
1. Barfield, B. J., R. I. Barnhisel, M. C. Hirschi and I. D. Moore. 1988. Compaction effects
on erosion of mine spoil and reconstructed topsoil. Transactions oftheASAE, 31(2):447-
452.
2. Hayes, J. C., B. J. Barfield and R. I. Barnhisel. 1984. Performance of grass filters under
laboratory and field conditions. Transactions oftheASAE, 27(5):1321-1331.
3. Lindley, M. R., B. J. Barfield, J. C. Ascough II, B. N. Wilson, and E. W. Stevens. 1998.
Hydraulic simulation techniques incorporated in the surface impoundment element of
WEPP. Applied Engineering in Agriculture, 14(3):249-256.
4. Williams, J.R. and A.D. Brendt. 1972. Sediment yield computed with universal
equation. Proceedings ASCE, 98(HY12):2087-2098.
5. Wilson, B. N., B. J. Barfield, A. D. Ward and I. D. Moore. 1984a. A hydrology and
sedimentology watershed model part I: Operational format and hydrologic component.
Transactions of the ASAE, 27(5): 1370-1377.
6. Wischmeier, W.H., C.B. Johnson, and B.V. Cross. 1971. A soil erodibility nomograph
for farmland and construction sites. J. Soil Water Conserv. 26:189-193.
Report
1. Barfield, B. J. and Hayes, J. C. 1992. Unpublished results of field evaluation of
sediment controls in South Carolina and Louisville, KY construction operations. J. C.
Hayes and Associates, Clemson, SC.
2. Barrett, M. E., Kearney, J. E., McCoy, T. G., Malina, J. F.,Charbeneau, R. J. and Ward,
G. H. 1995. An evaluation of the use and effectiveness of temporary sediment controls.
Technical Report CRWR 261, Center for Research in Water Resources, The University
of Texas at Austin, Austin, TX 78712.
3. Driscoll, E.D., D. DiToro, D. Gaboury, and P. Shelley. 1986. Methodology for analysis
of detention basins for control of urban runoff quality. Report no. EPA 440/5-87-01
(NTIS No. PB87-116562). U.S. Environmental Protection Agency, Washington, DC.
4. Elliott, W.J., A.M. Liebenow, J.M. Laflen, and K.D. Kohl. 1989. A Compendium of Soil
Erodibility Data From WEPP Cropland Field Erodibility Experiments 1987 and 88.
Ohio State University and USDA Agricultural Research Service. NSERL Report No. 3.
58
-------
5. EPA. 1993. Guidance Specifying Management Measures for Sources ofNonpoint
Pollution in Coastal Waters. EPA 840 B 92 002. U.S. Environmental Protection Agency,
Office of Water, Washington, DC.
6. EPA. 1999. Preliminary data summary for urban storm water management practices.
EPA-821-R99-012.
7. EPA. 2001. The National Water Quality Inventory: 2000 Report. EPA-841-R-02-001,
Washington, DC.
8. EPA. 2002. Development document for proposed effluent guidelines and standards for
the construction and development category. EPA-821-R-02-007.
9. EPA. 2002. Invitation only conference on sediment. Cincinnati, OH, July.
10. Knisel, W. G. editor. 1980. A field-scale model for chemicals, runoff, and erosion from
agricultural management systems. U. S. Department of Agriculture, Conservation
Research Report No. 26.
11. Natural Resources Conservation Service. 1969. NEHPart 630, Hydrology: Chapter 10-
-Estimation of Direct Runoff from Storm Rainfall.
12. Robichaud, P. R. and R. E. Brown. 2002. Silt Fences: An economical technique for
measuring hillslope soil erosion. USDA Forest Service, General Technical Report
RMRS-GTR-94, Rocky Mountain Research Station, Moscow, ID.
13 Smith, R., R. Alexander, and K. Lanfear. 1992. Stream water quality in the
conterminous US- Status and trends of selected indicators during the 1980s. USGS P-
2400. Washington, DC.
14. Smolen, M.D., D.W. Miller, L.C. Wyall, J. Lichthardt, and A.L. Lanier. 1988. Erosion
and Sediment Control Planning and Design Manual. North Carolina Sedimentation
Control Commission and North Carolina Department of Natural Resources and
Community Development, Raleigh, NC.
15. Transportation Research Board. 1980. Erosion control during highway construction.
Manual on principles and practices. Report 221. National Cooperative Highway
Research Program. Transportation Research Board, National Research Council,
Washington, DC.
Book or Book Chapter
1. Britton, S. L. 1999. Performance evaluation of silt fences for controlling sediment
release. M.S. thesis, Biosystems and Agricultural Engineering Dept, Oklahoma State
University, Stillwater.
2. Cross, C. 2003. New angles on silt fence. M.S. thesis, Biosystems and Agricultural
Engineering Dept., Oklahoma State University, Stillwater.
59
-------
3. Haan, C. T., B. J. Barfield, and J. C. Hayes. 1994. Design Hydrology and Sedimentology
for Small Catchments. Academic Press, New York.
4. Storm, D. E. 1991. Modeling Dynamic Rill Networks From Random Surfaces on
Moderate Slope. Doctoral Dissertation, Agricultural Engineering Department, University
of Kentucky, Lexington, KY 40506.
Proceedings
1. Barfield, B. J., J. C. Hayes, A. W. Fogle, and K. A. Kranzler. 1996. The SEDIMOT III
Model of Watershed Hydrology and Sedimentology. Proceedings of Sixth Federal
Interagency Sedimentation Conference, March.
2. Wishowski, J. M., Mamo, M., and Bubenzer, G. D. 1998. Trap efficiencies of filter
fabric fence. Paper No 982158, American Society of Agricultural Engineers, St. Joseph,
MI.
3. Robichaud, P. R, D. K. McCool, C. D. Pannkuk, R. E. Brown, and P. W. Mutch. 2000.
Trap efficiency of silt fences used in hillslope erosion studies. In: Ascough II and
Flanagan eds. Soil Erosion for the 21st Century: Proceedings of the International
Symposium. ASAE Pub. No. 701P0007, St. Joseph MI. January.
60
------- |