You are on page 1of 18

LIFESim 201

DAM AND LEVEE SAFETY RISK ASSESSMENT EVALUATION ROUTING


AND LIFE LOSS ESTIMATION USING LIFESIM

Woodrow Fields, P.E.
1

Chris Bahner, P.E., D. WRE
2

Jason Needham
3

Christopher R. Goodell, P.E., D. WRE
4


ABSTRACT

LIFESim is a modular, spatially-distributed, dynamic simulation system for estimating
potential life loss from dam and levee failure or non failure flood events that explicitly
considers the primary factors contributing to life loss in a flood situation. LIFESim is the
U.S. Army Corps of Engineers (USACE) most rigorous approach for estimating potential
life loss due to dam failure, and it can be used to provide inputs for dam safety risk
assessment. LIFESim considers detailed flood dynamics, loss of shelter, warning and
evacuation, and uses historically-based fatality rates to estimate life loss.

The Warning and Evacuation Module spatially redistributes the population at risk from
its initial distribution at the time that a warning is issued, to a new distribution with
assigned flood zone categories at the time of arrival of the flood. USACE is in the
process of improving the LIFESim evacuation-transportation process to account for
defining initial escape routes based on shortest travel time and allowing evacuees to turn
around if they come to a flooded road. USACE is also in the process of performing a
comparison study to demonstrate that LIFESim is appropriate for a breach and/or
overtopping scenario of a levee structure. In support of the comparison study, LIFESim
was applied to two highly urbanized areas.

This paper provides an overview of LIFESim, discusses the changes to the evacuation-
transportation process in LIFESim and their expected effects on life loss estimation, and
presents development and results of LIFESim models for the one of the urbanized areas.

INTRODUCTION

There are presently about 4,000 deficient dams in the United States that would require
about $12.5 billion dollars to repair and rehabilitate, and approximately 100,000 miles of
levees that have an unknown reliability (average age of 50 years) and an estimated cost of
$50 billion to repair and rehabilitate (ASCE, 2010). Addressing the aging infrastructure

1
Hydraulic Engineer, U.S. Army Corps of Engineers, Institute For Water Resources, Hydrologic
Engineering Center, 609 Second Street, Davis, CA 95616; ph: 530 756-1104;
Woodrow.L.Fields@usace.army.mil.
2
Senior Hydraulic Engineer, WEST Consultants, Inc., 2601 25th Street SE, Suite 450, Salem, OR 97302,
ph: 503-485-5409, cbahner@westconsultants.com.
3
Senior Consequence Specialist, U.S. Army Corps of Engineers, Risk Management Center, 609 Second
Street, Davis, CA 95616, ph: 503-756-1104, Jason.T.Needham@usace.army.mil.
4
Portland Office Manager/Senior Hydraulic Engineer, WEST Consultants, Inc., 103000 SW Greenburg
Rd., Suite 470, Portland, OR 97223, ph: 503-946-8536, cgoodell@westconsultants.com.
202 Innovative Dam and Levee Design and Construction
will be a challenging task to accomplish due to the sheer number of dams and levees that
need to be repaired, regulatory and environmental requirements, and funding limitation.
Prioritization should be an important component during the process to reduce the risk for
economic and social impacts, infrastructure damage, and loss of life associated with
potential future dam or levee failures.

One of the important components for assessing flood risk is to properly estimate the loss
of life from a dam or levee failure. Various empirical equations that are based on
population at risk (PAR) and the warning time have been developed to estimate the life
loss for a dam failure. McCelland and Bowles (2002) provide a detailed review of these
equations and identify several limitations with the empirical life-loss estimation
approach.

LIFESim was developed to overcome the limitation of the empirical life-loss estimation
approach. LIFESim is a spatially-distributed dynamic simulations modeling system for
estimating life-loss. LIFESim is currently being developed by the USACE. The
prototype application developed by Aboelata and Bowles (2005) was created for the
ArcView3 geographic information systems (GIS) platform. The first stages of the
LIFESim extension development by USACE involved the upgraded from AcView3 to the
ArcGIS ArcMap (Version 9.3) platform. Future development will include the
incorporation of additional methodology and/or improvements to the existing modules.

LIFESim has been applied to several dams under a range of failure and exposure
scenarios. The USACE Risk Management Center (RMC) in collaboration with the
Hydrologic Engineering Center (HEC) is in the process of performing a comparison
study to demonstrate that LIFESim is appropriate for a breach and/or overtopping
scenario of a levee structure. In support of the comparison study, LIFESim was applied
to two highly urbanized areas: (1) Saint Paul Flood Control Project in St. Paul, MN, and
(2) Sacramento River Flood Project (Natomas Basin) in Sacramento, CA. With the
incorporation of updates necessary to evaluate levee failures, LIFESim will become an
important tool for performing dam and levee safety risk assessment and for improving the
effectiveness of emergency action planning and responses.

LIFESIM OVERVIEW

LIFESim is a modular, spatially distributed, dynamic simulation system built around
databases. Each module exchanges data with other modules through the database, which
include various layers and tables containing information on population, structure types,
roads, warning systems, and time profiles of flood depth and velocity. Development of
LIFESim has been sponsored by the USACE and the Australian National Committee on
Large Dams (ANCOLD). LIFESim models are developed and run in the ArcGIS
environment. As presented in the LIFESim Users Manual (USACE, 2010), the
methodology implemented in LIFESim is based on research and development efforts of
Maged A. Aboelata and Dr. David Bowles of the Institute for Dam Safety Risk
Management of Utah State University and is described in the report, LIFESim: A Model
for Estimating Dam Failure Life Loss, Draft (Aboelata and Bowles, 2005).
LIFESim 203

A schematic of the LIFESim approach to life-loss estimation is shown in Figure 1.
LIFESim is comprised of three major modules. The following discussion related to each
of the modules is summarized from a paper prepared by Aboelata and Bowles (2005) that
contains a more detailed discussion of each module.



Figure 1. Schematic of Life-loss Estimation Approach in LIFESim (Aboelata and
Bowles, 2005).

Population at Risk (PAR)

LIFESim requires an estimate of the spatial and temporal distribution of population at the
time of the initiation of the first evacuation warning for a specific failure scenario. To
facilitate gathering these data, LIFESim uses readily available data. Specifically,
LIFESim was built to gather this information from the extensive database that
accompanies FEMAs HAZUS-MH software program (Aboelata and Bowles 2006). The
HAZUS database includes a polygon shapefile that delineates census blocks for an area
as well as the population and building characteristics for each of those census blocks.

Variations in the number of people within the flooded area throughout the day can have a
significant effect on loss of life. Moreover, the activities that people are engaged in can
affect the efficiency at which the warning spreads through an area. Accordingly,
LIFESim is designed to take into account both 1) the distribution of people among census
blocks throughout the inundated area based on time of day and 2) estimates in the
percentage of population involved in various activity types including: at home, outdoors,
working/shopping and in transit.
204 Innovative Dam and Levee Design and Construction
To determine the number of people in a census block at any given time, LIFESim pulls
relevant data from the HAZUS-MH database. The HAZUS-MH database contains three
time-of-day activity distributions. These distributions are as follows: 0200, representing
night time; 1400, representing day time; and 1700, representing commuting time.
LIFESim takes the data and uses relationships to calculate the number of people in each
occupancy class on the census block level.

Variables used in distributing people are as follows:

Census block population taken from census data stored in HAZUS-MH database
Daytime residential population inferred from census data
Nighttime residential population inferred from census data
Number of people commuting inferred from census data
Number of people employed in the commercial sector
Number of people employed in the industrial sector
Number of students in grade schools (K-12)
Number of students on college and university campuses in the census tract
Number of people staying in hotels in the census tract
A factor representing the proportion of commuters using automobiles, inferred from
profile of the community (0.60 for dense urban, 0.80 for less dense urban or
suburban, and 0.85 for rural). The HAZUS-MH default value is 0.80.
Number of regional residents who do not live in the study area, visiting the census
tract for shopping and entertainment. The HAZUS-MH default value is zero.

LIFESim interpolates the estimates obtained by applying the HAZUS-MH population
data for the three HAZUS-MH time-of-day distributions to obtain estimates for every
two-hour interval throughout a 24-hour period. Population estimates for each of the four
LIFESim activity types are then computed using the mapping shown in Table 1.

Table 1. Mapping of HAZUS-MH occupancy classes to LIFESim activity types.
LIFESim activity type HAZUS occupancy class
Residential-indoors Residential-indoors
Hotels-indoors
Outdoors Residential-outdoors
Hotels-outdoors
Working/shopping Commercial-indoors
Commercial-outdoors
Industrial-indoors
industrial-outdoors
Education-indoors
Education-outdoors
In transit Commuting-own car
Commuting-public transportation

LIFESim 205
Finally, population is distributed among activity types using a set of activity factors.
These values are based on research from Rogers and Sorensen (1988). Their research
described how a given warning system has a penetration capability that can be
distinguished for the following five fundamental locations or activities:

1) Home asleep
2) Indoors at home or in the neighborhood
3) Outdoors in the neighborhood
4) In transit
5) Working or shopping

They also added two other activities: watching television and listening to radio,
which override the first set of five locations or activities.

Loss of Shelter

The Loss of Shelter Module simulates the exposure of people in buildings during a flood
event as a result of structure damage, building submergence, and toppling of people in
partially damaged buildings. The module utilizes historical fatality-rate probability
distribution for three distinct flood zones: (1) Chance Zone, which is the zone where
victims are impacted by the floodwaters and survival is depended mainly on chance
(fatality rate of 50 to 100% with an average of 90%); (2) Compromised Zone, which is
the zone where victims are exposed to floodwaters as a result of damage to the shelter by
the flooding (fatality rate of 0 to 50% with an average of 10%); and (3) Safe Zone, which
is the zone that is either dry or consists of shallow, tranquil floodwaters (virtually no life
loss in this zone).

LIFESim includes two different methodologies for building damage criteria: (1) USACE,
or (2) RESCDAM. The USACE criteria consists of a set of collapse curves based on
depth and flow velocity for one, two, and three story buildings of four different types of
buildings. The RESCDAM criteria are based on a detailed study completed in Finland,
and it defines when a wood-frame or masonry, concrete, brick building would be partially
or totally damaged based on depth and flow velocity data.

Warning and Evacuation

The Warning and Evacuation Module simulates the spatial redistribution of the
population at risk following the intuition of being warned. This is accomplished through
simulation of the warning dissemination, mobilization, and evacuation-transportation
processes. Current improvements in the LIFESim evacuation-transportation process
account for defining initial escape routes based on shortest travel time and allowing
evacuees to turn around if they come to a flooded road. The following will discuss in
more detail the warning, mobilization, and evacuation process.

The warning initiation time is the time at which an evacuation warning is first issued to
the PAR. It is defined to be positive if the warning is issued after failure occurs, or to be
206 Innovative Dam and Levee Design and Construction
negative if the warning is issued before failure occurs. The warning module start time at
time step zero is the earliest warning initiation time. Any emergency planning zone
(EPZ) with a later initiation time is offset accordingly. The rate at which the warning is
received throughout an area is represented in LIFESim using a warning diffusion curve,
which is the cumulative percentage of the PAR that receives the warning message versus
time where time 0 is the warning initiation time.

After receiving the warning message, people who are willing and able to leave will
prepare to leave. The rate of mobilization is represented in LIFESim using a
mobilization curve, which is a cumulative percentage of the warned PAR that starts
moving away from the area of potential flooding towards safe destinations. A
mobilization curve represents two important pieces of information: (1) how long it takes
people after they received a warning to leave their home, and (2) what percentage of the
population will not mobilize (1 minus the maximum mobilization %). Typically a
mobilization curve will not reach 100% until enough time has passed to allow emergency
responders to physically enter every home and remove people that are either unable or
unwilling to mobilize on their own.

The evacuation-transportation process commences with mobilization and ends with either
clearance of the flooding area or entrapment if the evacuation route becomes blocked by
flooding. People who clear the flooding area are assigned to a safe flood zone and
people who are trapped on the road are assigned to a flood zone that depends on their
mode of evacuation and the most severe flooding conditions for the event. Three modes
of evacuation are included in LIFESim: cars, sports utility vehicles (SUVs) and
pedestrians.

The Greenshield (1935) transportation model is used in LIFESim to represent the effects
of traffic density and road capacity on vehicle speed. The original model was modified to
represent congestion and traffic jams, as described in Aboelata and Bowles (2005), by
introducing a minimum stop-and-go speed (Vjam) if the jam density (Djam) for a road
class is exceeded. Each road class is assigned default values of the number of lanes, free
flow speed (ffs), Djam and Vjam based on the Highway Capacity Manual (HCM) (TRB
2000), although these can be overridden if more detailed information is available for the
road system. In the case where traffic jams occur, a minimum stop-and-go speed is used
when jam density is exceeded. A road segment is determined to be blocked when the
vehicle stability criteria has been exceeded.

A road network in the form of a GIS polyline shapefile is provided in which each
segment of the network contains information such as road category, segment length,
number of lanes, and interconnectivity. The road segment end points must be snapped
together so that the shortest path algorithm knows that they are connected. LifeSim also
has the capability for the user to create a driveway segment which draws a line from
the census block centroid to the nearest road segment. The driveway segment road
category is user defined at the time of creation.

LIFESim 207
A GIS point shapefile represents a set of safe destinations for evacuees to go. LIFESim
models the movement of people to destination points using the shortest travel time
available. The destination points are not required to be snapped to a road segment, they
can be placed directly on the safe location. People who reach a safe destination are
considered as the cleared group.

During each time step at the user defined interval t, groups move as far as the model
allows until a destination point is reached. If a flooded road is encountered, a new route
is determined. The population group then keeps the flooded road in memory so that it
will not try to evacuate through the road again. If no new route can be found, the
population group is stranded. The evacuation routes are determined by computing the
shortest travel time as if all the roads were capable of free flow speed. The shortest path
algorithm that is used to compute the evacuation routes is the Dijkstra with Approximate
Buckets method (Cherkassky et al. 1993). This algorithm was chosen for its stability and
speed. Also, it was chosen based on a comparison effort of various shortest path
algorithms (Zhan and Noon, 1998).

Loss of Life

The final step in LIFESim is the estimation of loss of life. Previously described modules
simulate the spatial redistribution of people existing within the study area through the
processes of warning and evacuation and assign loss-of-shelter category/flood zones
based on the effect of flood water on the buildings, vehicles and pedestrians throughout
the study region. These results are combined in the Loss-of-Life (LOL) Module with the
probability distribution of fatality rates for each loss-of-shelter category/flood zone to
obtain estimates of the expected number of fatalities within the study area. For this
analysis, the expected value (mean) of the fatality rate distributions developed by
McClelland and Bowles (2000), updated Aboelata et al (2003) and displayed in Figure 2
were applied. The mean values for the safe, compromised and chance flood zones are
0.0002, 0.1200, and 0.9145, respectively.

208 Innovative Dam and Levee Design and Construction
COMBINED DATA
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Rel ati ve Frequency of Exceedance
P
r
o
p
o
r
t
i
o
n
a
l

L
i
f
e

L
o
s
s
,

P
r
(
z
o
n
e
)
Prcz Prcoz Prsz

Figure 2. Probability distributions for fatality rates for each flood zone.

LIFESIM APPLICATION

LIFESim was applied to two highly urbanized areas to evaluate the applicability of the
program to areas protected by levees: (1) Saint Paul Flood Control Project in St. Paul,
MN, and (2) Sacramento River Flood Project (Natomas Basin) in Sacramento, CA.
However for this paper, the necessary steps for the development of a LIFESim model and
results from the LIFESim simulation are presented only for the Natomas Basin in
Sacramento, CA.

Study Site

The Natomas Basin is shown in Figure 3. It is located just north of the confluence of the
American and Sacramento Rivers. The Natomas Basin has a total protected area of about
83.2 mi
2
, and it includes portions of the city of Sacramento, Sacramento County, and
Sutter County. The Natomas Basin is bordered on the south by the American River, on
the west by the Sacramento River, on the north by the Natomas Cross Canal (NCC), and
on the east by the Pleasant Grove Creek Canal (PGCC) and the Natomas East Main
Drainage Canal (NEMDC).

The Natomas Basis is protected from high flood flows by an interconnected perimeter
levee system. The levees are divided into four major units: (1) Levee Unit 1 is located
along the east bank of the Sacramento River for a total length of 18.6 miles, (2) Levee
Unit 2 is located along the north bank of the American River for a total length of 2.3
miles, (3) Levee Unit 3 is located along the west bank of the NEMDC and Pleasant
Grove Creek Canal for a total length of 17.3 miles, and (4) Levee Unit 4 is located along
the south bank of the NCC for a total length of 4.4 miles. This levee system was
originally created to promote agricultural development. However, the Natomas Basin
presently contains three major public transportation facilities: (1) Interstate 5 (I-5), (2)

LIFESim 209

Figure 3. Natomas Basin

Interstate 80 (I-80), and (3) State Route 99 (SR-99); the Sacramento International
Airport; the Arco Arena; and several commercial and residential developments. About
30% of the basin consists of developed urban uses, mostly in the southern portion of the
basin in the city of Sacramento. The remaining portion of the basin is in some form of
developed agricultural or open space.

Model Development

LIFESim was utilized to evaluate various breach scenarios at three different locations.
The locations considered are shown in Figure 3, and consist of Sacramento River near the
Cross Canal confluence (northern location), Sacramento River near the Sacramento Weir
(middle location), and lower reach of the American River/NEMDC (southern location).
The breach scenarios considered included Sunny Day Failure, Expected Failure from
210 Innovative Dam and Levee Design and Construction
Overtopping, and No Failure (Strong Levee). The three breach scenarios were
considered for the northern and southern breach locations. The Sunny Day Failure
scenario was the only failure scenario considered for the middle location since the levee
at this location would not be overtopped during the 0.2 annual chance flood event and
significant overtopping of the other levee segments would be occur when the levee at this
location would be overtopped.

The general steps and procedure in developing the LIFESim models are presented in the
following paragraphs.

Map Projection. As previously discussed, LIFESim runs in an ArcGIS environment. So,
it is important that the various shapefiles and raster files have the same projection. The
hydraulics for the Natomas Basin was estimated using HEC-RAS and FLO-2D. Both of
these models were developed using data with a horizontal datum of the NAD 1983
California II State Plane, feet. Therefore, this datum was utilized for the LIFESim
models of the Natomas Basin.

HAZUS Database. Populations and building information utilized by the LIFESim
program is based on information available from various databases developed as part of
the Federal Emergency Management Agencys Hazards U.S. Multi-Hazard (HAZUS-
MH) application (FEMA, 2003). The HAZUS-MH application and supporting database
files were obtained from FEMA. The tables and database files used from the HAZUS
system include: (1) hzDemographicsB table, (2) hzBldgCountOccupB table, and (3)
hzCensusBlock table from the bndrygbds.mdb database; and (4) hzGenBldgScheme from
the MSH.mdb database.

Census Block Layer. A census block polygon shapefile was developed from the Census
Blocks feature class in the bndrygbs.mdb HAZUS-MH database. The shapefile was
defined to contain only the polygons that are inundated for the maximum flooding
conditions. Also, the attribute table was revised as follows: (1) a field named Block_id
was added to the table and populated with values in the modified CensusBlock field in
the HAZUS-MH database, (2) a field for the percent of elders in each census block was
added and populated using information contained in the hzDemogrphaicsB table of the
bndrygbs.mdb HAZUS-MH database, and (3) a field for the emergency planning zone
(EPZ) was added and populated with the same value for all census block to represent a
single warning and evaluation scenario. Presently, no information has been obtained
related the Emergency Action Plan (EAP) for the Natomas Basin.

LIFESim requires a point shapefile that represents the centroid of each census block.
This shapefile was created using the Generate Centroid tool available as part of the
spatial preprocessing capabilities of the LIFESim software.

Evacuation Destination Layer. LIFESim requires a point shapefile that represents the
locations to which people will evacuate. The evacuation destination locations are shown
in Figure 4, and they were defined to be at the end of most of the roads entering and
leaving the basin.
LIFESim 211
Road Layer. The existing roads within the Natomas Basin are shown in Figure 4. The
roads layer shapefile representing the transportation network was provided by the U.S.
Army Corps of Engineers, Hydrologic Engineering Center (HEC). The shapefile
included the necessary field to link the Census Feature Class Codes for each road
element. The weight of the lines for the roads in the figure corresponds to the capacity of
the road system with thicker lines having a higher vehicle capacity and maximum speed
limit.

LIFESim requires that the road lines extend to the center of each census block. This was
accomplished using Add Driveway tool available as part of the spatial preprocessing
capabilities of the LIFESim software.

First Floor Height Grid. Information related to the first floor elevations is provided as a
raster dataset that must have the same cell size as the raster datasets of the flood water
depth and velocity. The first floor raster was developed using the FLO-2D grid, which
was used to define the flood water depth and velocity raster datasets, and assuming that
the first floor of the structures are 3 feet above the existing ground.

Flood Profile Data. LIFESim requires a time series of water depth and velocity data in
raster format. The time series data must cover the time of evacuation being modeled, and
it can be defined using 30 time steps. The depth and velocity raster files must be named
using the convention of <prefix><sequence number>, i.e., names for the depth are dp001
and dp002 and for the velocity are vp001 and vp002; and they must be located in the
same directory.

The raster dataset was defined using the unsteady flow HEC-RAS and FLO-2D models
provided by Sacramento District (SPK) of the U.S. Army Corps of Engineers. The
unsteady flow HEC-RAS model was used to define the inflow hydrograph to the FLO-2D
model. The HEC-RAS model was developed by SPK as part of the American River
Common Features GRR study (USACE, 2002). The provided model consisted of only
one plan for the 0.002 exceedance probability flood event. The model covers the lower
reach of the Sacramento River, from Colusa at the upstream end to the upper portion of
the Sacramento-San Joaquin Delta at the downstream end. The model also includes
major tributaries (Feather, Bear, Yuba, and American Rivers), several of the smaller
tributaries (Putah and Cache), and various interconnecting waterways. The upstream
boundaries consist of an unsteady flow hydrograph; the internal boundaries consist of
either a steady or unsteady flow hydrograph or gate control structure, and the downstream
boundaries consist of an unsteady stage hydrograph. The downstream boundaries are
based on the stage hydrograph measured during the 1997 flood event and adjusted using
the results of frequency analysis on the tidal data located near each of the boundary
locations.


212 Innovative Dam and Levee Design and Construction

Figure 4. Evacuation Destination Point and Roads for Natomas Basin


The HEC-RAS model is highly complex and includes about 96 reaches, 3,930 cross
sections, 159 storage areas, 582 lateral structures, 186 storage connection features, and
146 bridge and/or culvert structures. The Mannings n-values for the main channel range
from 0.035 to 0.06 and 0.035 to 0.30 for the overbank areas. The weir coefficients of the
lateral weir structures range from 1.4 to 3.0. There is also one diversion structure with 48
gates that is located on the right bank of the Sacramento River about 2,700 feet upstream
of the Highway 80 Bridge.

The levees are simulated in the model using lateral structures. Therefore, the levee
breach tool available in HEC-RAS was used to breach the levee. Piping failure mode
was utilized for the Sunny Day failure scenarios, while the overtopping failure mode was
considered for the Overtopping failure scenarios. Each breach was assumed to be 500
LIFESim 213
feet wide and with vertical side slopes that is formed over a duration of 1 hour. Also the
breach weir coefficient was assumed to be 2.6 and final bottom elevation set by the
interior ground elevation where the breach occurs. The resulting hydrograph was placed
into Excel and converted to an ASCII format for easy input into the FLO-2D model.

FLO-2D software (FLO-2D, 2009) was utilized to define the hydraulics dataset of the
Natomas Basin for the LIFESim application. FLO-2D is a horizontal 2-D flood routing
model developed by Dr. Jim OBrien. It numerically routes a flood hydrograph over a
computational domain while predicting the area of inundation and simulating floodwave
attenuation. It is a volume conservation model that moves the flood volume from
upstream to downstream on a series of computational cells (elements). Floodwave
progression over the flow domain is controlled by topography and resistance to flow.
Flood routing in two dimensions is accomplished through an explicit numerical
integration of the equations of motion and the conservation of fluid volume. The model
can solve either the diffusive wave equation (neglecting the acceleration terms in the
momentum equations) suitable for simple overland flow on a mild slope, or the full
dynamic wave equation for simulating complex flow hydraulics.

The FLO-2D model provided by SPK consisted of 36,475 square elements that are 400
feet by 400 feet (210 mi
2
). The model covers the entire Natomas Basin (about 83 mi
2
)
and storage areas of the NEMDC, the area between the Sacramento River and Yolo
Bypass, and the area between the American River and Arcade Creek. The FLO-2D
model included levee, inflow and outflow elements; Area Reduction Factors (ARF) to
reflect blockage associated with buildings within urbanized areas, and a channel reach to
represent the NEMDC.

New FLO-2D models were developed for the various breach locations and failure
scenarios by revising the inflow conditions to the model. The outflow hydrographs from
the HEC-RAS were reflected as an inflow hydrograph to FLO-2D. The FLO-2D
software has the capability (Mapper program) to create shapefiles for specific time
increments. The process, however, is fairly time intensive. Therefore, a macro was
developed in Excel to extract the FLO-2D output at selected time increments and create
ASCII files in a format that can easily be converted to a shapefile and then into a raster
format in ArcGIS.

Model Setup. Prior to running LIFESim, the input parameters need to be defined in the
model setup in terms of the initial settings, spatial/population data, warning/evacuation,
and depth/velocity data. Brief information related to the model setup is presented as
follows.

Initial Settings. Some of the basic information is defined in the initial settings dialog,
including the model output directory, time of day options, building damage criteria, and
spatial extents. As previously mentioned, there are two different types of building
damage criteria available within LIFESim: (1) USACE, or (2) RESCDAM. The USACE
method was selected for the evaluation of the Natomas Basin. The spatial extents
specified in the initial setting dialog define the area of the computation limits, and it
214 Innovative Dam and Levee Design and Construction
should be large enough to contain the census block, evaluation destinations and roads to
the destinations.

Spatial and Population Data. The Spatial and Population Data dialog defines what spatial
data layers and population data files that will be used in the LIFESim analysis.

Warning and Evacuation Setup. The Warning and Evaluation Setup dialog consist of two
tabs. The first tab is for defining the parameters related to the Warning/Mobilization.
The type of warning system, warning issuance, mobilization curve, and population
activity distribution is defined for each EPZ. There are eight types of warning systems
available: (1) Sirens; (2) Tone Alert Radios; (3) Emergency Broadcast System (EBS); (4)
EBS and Sirens; (5) EBS and Auto-Dial Telephones (Landlines) (ADT); (6) EBS, Siren,
and ADT; (7) EBS, Sirens, and ADT (Landlines and Mobile); and (8) User Define. It
was assumed that the Natomas Basin had only one EPZ with an EBS warning system,
and the warning issuance, which defines the time in minutes between when the warning
is issued and dam/levee failure that the warning issuance, would be instantaneous (0
minutes). The mobilization curve defines the relationship between the percent mobilized
over time. There are three pre-defined mobilization curves to choose from (Default,
Below Average, and Above Average) plus one user defined option. The default curve
was utilized for the Natomas Basin. Finally, the population activity distribution is
utilized to define the percentage of the population engaged in each of seven activities in
2-hour increments over a 24-hour time period. As in the mobilization curve, the default
distribution was assumed for the Natomas Basin.

The second tab is for defining the control settings of the evacuation model including the
model duration, vehicle characteristics, and additional evacuation parameters. The model
time includes the computation time step and the total simulation time in minutes. The
computational time step should range between 1 to 10 minutes with a larger time step
being considered for larger areas. The total simulation time defines the total time in
minutes to perform the warning and evacuation process in the model. The Natomas
Basin was evaluated with a computational time step of 10 minutes and a total simulation
time varied dependent on the breach conditions. The vehicle characteristics includes: (1)
the vehicle occupancy rate, which is the average number of people per vehicle (default of
3); (2) the fraction of people in vehicles, which is the fraction of people that will evacuate
using vehicle rather than on foot (default of 1); and (3) the fraction of people in cars
versus SUVs/trucks (default value of 0.5). Finally, the additional evacuation parameters
dialog can be utilized to revise: (1) the road characteristics, which includes information of
the number of lanes, free flow speed in mi/hr, and jam density; (2) the stability factors for
humans, cars and SUVs (set as depth times velocity factor, and default value of 1, 1.35,
and 0.9 for humans, cars, and SUVs, respectively); and (3) speed adjustment factors,
which are used to scale the model pedestrian speed of 4 mi/hr and the vehicle free flow
speed defined in the road characteristic table (default values of 1). For the LIFESim
evaluation of the Natomas Basin Default values were assumed for the vehicle
characteristics and additional evacuation parameters.

LIFESim 215
Depth/Velocity Data. The Depth and Velocity Profile Information dialog is utilized to
define the location of the raster depth and velocity files, what the starting depth and
velocity raster file, number of profiles and time in hours associated with each profile, and
time in hours of the dam/levee failure with respect to the hydraulic model start.

Model Results

The LIFESim results for the time of failure that resulted in the greatest loss of life are
summarized in Table 2.

Table 2. Summary of LifeSim Results.
Breach
Scenario Variable
Northern
Location
(Sacramento
River near Cross
Canal)
Southwest
Location
(Sacramento
near Sac Weir)
Southern
Location
(American River
and NEMDC
Combined
Reach)
Sunny Day
Time of Failure
18:00 18:00 18:00
Cleared
44,756 41,842 42,289
Survived
812 887 2,973
Loss of Life
104 35 397
Expected
(Overtopping)
Failure
Time of Failure
18:00 - 0:00 4:00
(1)

Cleared
45,283 - 36,633
Survived
853 - 4,308
Loss of Life
71 - 476
Strong
Levees
Time of Failure
18:00 - 0:00 4:00
(1)

Cleared
3,735 - 36,733
Survived
76 - 3,285
Loss of Life
0 - 296
Notes:
1. For the Southern Location, the results were the same for the time of failure occurring at 0:00, 2:00, and 4:00
for the Expected (Overtopping) and Strong Levee breach scenarios.
2. The Expected (Overtopping) and Strong Levee breach scenarios were not considered for the Southwest
Location.

A review of the LIFESim results indicate the following: (1) the greatest loss of life would
occur for the breach and/or overtopping conditions occur at the southern location, which
is anticipated since it the closest location to a populated area; (2) the breach condition
with the greatest loss of life for the southern levee location would be from a Overtopping
failure; (3) the breach condition with the greatest loss of life for the northern levee
location would be from a Sunny Day failure; (4) the time of failure that results in the
greatest loss of life is either at the early morning or early evening hours; (5) no lives will
be lost if the levee adjacent to the Sacramento River is protected against overtopping
failure; and (6) there will be about a 38 percent reduction in the potential loss of lives if
the levee adjacent to the American river and NEMDC are protected against overtopping
failure.
216 Innovative Dam and Levee Design and Construction
Based on a quick sensitivity analysis of the warning time for the Sunny Day failure
condition at the southern levee, a 30 percent reduction in loss of life will occur if the
warning is issued 30 minutes prior to the failure and a 40 percent increase in loss of life
will occur if the warning is issued 30 minutes after the failure.

CONCLUSION

Due to the current state of several dams and levees within the United States, there will be
an increasing demand and necessity for dam and levee risk assessments, improving the
effectiveness of emergency planning and responses to dam or levee failures, and
prioritizing the order of what dams and levees should be rehabilitated. Thus, a credible
estimation of life-loss from dam or levee failures is important. LIFESim has been
developed to meet this need. It has been formulated using the important processes that
can affect life loss, addresses many of limitations associated with existing empirical
approaches, utilizes readily-available data sources, and requires only a reasonable level of
effort to implement (Aboelata and Bowles, 2005). It is a fairly new tool that has been
successfully applied to estimate life-loss for several dams under a range of failure and
exposure scenarios, and recently to levee structures. There is on-going work to improve
the capabilities of the tool, such as the evacuation-transportation process discussed in this
paper, and USACE is committed in making LIFESim a powerful and easily accessible
tool.

Some general conclusions related to the application of LIFESim to the Natomas Basin
include: (1) the model run time could be extremely long (several days) for large interior
areas with failure that occur near populated areas; (2) the recommended upper limit of
simulation time (24 hours) for the warning and evacuation module could prevent the use
of the program for large interior areas where it could take several days for flooding of the
interior area; and (3) LIFESim relies on the HAZUS data, which is readily available but
could be outdated or contain erroneous information.

REFERENCES

Aboelata, M., and D. S. Bowles, 2008. LIFESim: A Tool for Estimating and Reducing
Life-Loss Resulting from Dam and Levee Failures Proceedings of the Association of
State Dam Safety Officials Dam Safety 2008 Conference, Indian Wells, CA.

Aboelata, M., and D. S. Bowles, 2005. LIFESim: A Model for Estimating Dam Failure
Life Loss Draft Report to Institute for Water Resources, US Army Corps of Engineers
and Australian National Committee on Large Dams.

Aboelata, M., D.S. Bowles and D.M. McClelland, 2004b. A Model for Estimating Dam
Failure Life Loss. ANCOLD Bulletin 127:43-62. August.

Aboelata, M., D.S. Bowles and D.M. McClelland, 2003. Life-loss Estimation for Floods
including Dam Failure. GIS Model for Estimating Dam Failure Life Loss. In Y.Y.
Haimes and D.A. Moser, (Eds.), American Society of Civil Engineers.
LIFESim 217
American Socity of Civil Egineers. 2010. Report Card for Americans Infrastructure.

FEMA (Federal Emergency Management Agency), 2003. HAZUS-MH flood technical
manual. Department of Homeland Security, Emergency Preparedness and Response
Directorate, FEMA, Mitigation Division, Washington, D.C.

FLO-2D Software, Inc., 2009 (January). FLO-2D Users Manual for Version 2009.06.

Hydrologic Engineering Center, 2010. HEC-RAS, River Analysis System Users
Manual. U.S. Army Corps of Engineers, Hydrologic Engineering Center, Davis,
California.

McClelland, D.M., and D.S. Bowles, 2000. Estimating Life Loss for Dam Safety and
Risk Assessment: Lessons from Case Histories. In Proceedings of the 2000 Annual
USCOLD Conference, U.S. Society on Dams (formerly U.S. Committee on Large
Dams), Denver, CO.

McClelland, D.M., and D.S. Bowles, 2002. Estimating Life Loss for Dam Safety Risk
Assessment - a Review and New Approach. Institute for Water Resources, U.S. Army
Corps of Engineers, Alexandria, VA.

U.S. Army Corps of Engineers, 2010. Draft LIFESim Extension Version 1.2 for ArcGIS
Users Manual. U.S. Army Corps of Engineers, Hydrologic Engineering Center, Davis,
California.

U.S. Army Corps of Engineers, 2002. Sacramento and San Joaquin River Basins
Comprehensive Study, Sacramento District.

You might also like