INTRODUCTION
Sediment deposition is the principal problem affecting the useful life of reservoirs. When a river enters a reservoir the flow velocity decreases and the sediment load begins to deposit. The bed load and coarse fraction of the suspended load are deposited immediately to form delta deposits, while fine sediments with lower fall velocity are transported deeper into the reservoir (Arman et al., 2009). Most of sediments are transported within reservoir to the point of deposition by three processes; transport of coarse material as bed load along the top set delta deposits, transport of fine sediment in density currents and also as nonstratified flow. Accumulation of these sediments reduces the useful life of storage over time. Usually the concept of the reservoir halftime is the time required to infill half the original capacity is an indicator of the life of conventional storage reservoir. Because the efficiency of sediment trapping declines as reservoir capacity is reduced, the halftime does not represent half the time required to lose all storage reduced (Morris and Fan, 1998).
Computation of the storage useful life requires the knowledge of sediment trapping
efficiency, the sediment unit weight or bulk dry density, the incoming flow
and sediment. Unit weight, specific weight and bulk density are all used to
express the dry weight per unit volume of a bulk sediment sample. The bulk density
of sediment deposited is estimated in twosteps processes: first the bulk density
of sediment in the first year of deposition is estimated as initial bulk density
and in second step the effect of compaction with time is computed. Since, the
void spaces between coarse grains are large enough, the water can freely escape
through the grains. As a result coarse sediments achieve essentially final density
as soon as they are deposited. Silt and clay, however, settle initially into
a loose matrix with inter particle bridging resulting in a large volume of small
water filled voids. When additional layer of sediment is deposited on the top
of the previous layer, additional pressure is applied and as a result the bridges
can collapse. This causes the bulk density of lower layer to increase over time.
The rate of increase is much higher at the beginning and then the rate decreases
(SamadiBoroujeni et al., 2005).
The process of sediment consolidation has been studied by Bloomquist and Townsend (1984). They found that the consolidation process can be separated into four stages. The first which occurs during the initial hours of the process is defined as stabilization agglomeration. In the second stage, which is called the sedimentation stage the void ratio may reach to a meaningful value of 5. The time span required for this process is approximately one day (24 h). In the third stage, the density of the sediments increases to such an extent that the settling of the particles effect one another considerably, yet the accumulated sediments have not cohered into a solid block. This stage is best known as the hindered settling stage. In the experiments that were carried out, the void ratio parameter was between 5 to 3. In the fourth stage the sediments consolidate and the pressure upon them gradually increases and the void ratio decreases over time. This stage is known as the selfweight consolidation stage (Bloomquist and Townsend, 1984).
Since, field investigation of reservoir sedimentation is expensive, application of the empirical equation with limited number of laboratory experiments are usual in practice. However, where accurate investigation is required, field investigation is a must. Field investigation in 13 small flood retention ponds in central Belgium by Verstraeten and Poesen (2001) showed that dry sediment bulk density varies not only between the selected ponds (0.781.35 t m^{3}) but also within individual ponds (coefficient of variation at 95% ranges from 7 to 80%). They pointed that the existing empirical relations are not a reliable predictor for the observed dry bulk densities because these relations are primarily based on sediment texture. Also, in a research on Grenada Lake, bulk density for 47 continuous sediment cores, ranging in length from 0.55 to 2.55 m were measured (Bennett and Rhoton, 2003). Due to low rate of deposition, sediment layers were thin. For example, sediments in depth of 0.5 and 0.9 m were 30 and 48 years old, respectively. The texture of sediments was 25 clay, 68 silt and 8% sand. The bulk density of sediments was measured to be 850, 1000, 1100 for depths 0, 0.7, 0.9 m and ages 1, 38, 48 years, respectively.
Despite having many relationships for predicting the bulk density of deposited sediment in reservoirs, nothing is done to evaluate the accuracy of these methods, in particular for reservoirs with high rate of deposition. This study was conducted to find out accurate estimation of bulk density, texture and compaction of sediments deposited behind Dez dam wall. This also allowed evaluation of most common empirical relationships in prediction of bulk density through comparison of the results.
MATERIALS AND METHODS
Theory
Due to complexity of sedimentation and consolidation process, empirical
relationships have been developed over the past years. Lara and Pemberton (1963)
developed an empirical method for estimating the initial specific weight of
sediment deposits based on the analysis of some 1300 sample from reservoirs
in the following form:
W_{1}
= W_{C}P_{C}+W_{M}P_{M}+W_{S}P_{S}  (1) 
where, W_{1} stands for the initial bulk dry density; P_{C}, P_{M} and P_{S} represents the percentages of clay, silt and sand, respectively and W_{C}, W_{M} and W_{S} are the initial bulk dry density for clay, silt and sand, respectively, in which their values depend on the reservoir operation method (Table 1). The initial bulk dry density can be obtained by the following empirical procedure:
Table 1: 
Coefficient B and initial bulk dry density values (kg m^{3})
for consolidation calculation 

• 
Dividing the sample into size groups and weight each size
group 
• 
Mixing each size group with water in separate calibrated container
and waiting until the particles are being settled 
• 
Measuring volume of the deposited materials, the initial bulk
dry density for each group, then for whole sample can be calculated using
Eq. 1 
It should be noted that for a clay soil, parameters of P_{M} and P_{S} would be considered as zero. In computing the average compaction over a period of time, each year’s sediment deposits will have a different compaction time. The average density of all sediment deposited during t years of consolidation may be computed using the equation presented by Miller (1953):
where, W_{t} is the average bulk dry density after t years of consolidation, W_{1} is the initial bulk dry density and B is constant as given in Table 1. Miller (1953) assumed that sediment accumulation occurs at a constant rate every year.
Lane and Koelzer (1943) presented an empirical formula for the densitytime relationship, which takes into account the grain size of the sediment and the method of operating the reservoir:
where, W is bulk dry density of a deposit with an age of t years; W_{1} is the initial bulk dry density, usually taken to be the value after one year of consolidation; and B is constant. Both W_{1} and B are functions of sediment size and they are defined for different operational condition.
Parameter values are given in Table 1. For sediment containing more than one size class, a weighted value for the coefficient B should be computed.
Dez Dam
The Dez dam is located in the Zagros Mountains in the Southwest Iran and was
constructed in 1963, with the height of 203 m. An underground powerhouse contains
eight 65 MW units for a total installed capacity of 520 MW which has generated
an average of 2400 GWh year^{1} energy production over an operating
period of 45 years. The minimum reservoir operation level is with the elevation
of 300 m and the maximum controlled level is with the elevation of 352 m. The
original reservoir volume was 3315 million m^{3}, but the results of
bathymetry survey in 2003 shows that the storage volume of the reservoir has
been reduced to 2600 million m^{3} by the sedimentation which corresponded
to a volume loss of about 19%. The incoming sediment has formed a delta which
has proceeded about halfway down the 50 km long reservoir as well as bottom
set stratified beds near the dam (Samadi Boroujeni et al., 2005). The
sediment level near the dam has risen at a rate of about 2 m year^{1}
over the operating period (Fig. 1, 2). Hence,
the reservoir bed at the face of the power intakes has increased from an original
elevation of 180 to 260 m in 2007, which is only 10 m below the invert level
of the power intake at the elevation of 270 m. There is now concern that sediment
will begin to be drawn into the power tunnels within a decade with potential
damaging effects on the turbine runners and other water passage elements.

Fig. 1: 
Longitude profiles of Dez reservoir talweg versus years of
dam operation 

Fig. 2: 
Increase of bed elevation (near Dez dam) versus time based
on the reservoir bathymetry maps 
In addition, this deposition adjacent to the dam is now about 37 m above the
lowlevel outlets within the dam body. This situation has an impact on the physical
operations at the dam, including the irrigation outlets, power generation and
reservoir operation (Samadi and Galay, 2005).
In order to decrease sediment level in front of the power intakes, operator of the Dez dam, Khusestan Water and Power Authority (KWPA), has a plan to open the low level outlet (i.e., irrigation outlets) for discharging the sediment deposited in front of the power intake as an operational program. In this regards, it is essential to determine the physical properties of the sediment deposited in front of the power intakes, so, KWPA arranged a field investigation for sampling of sediments by digging a deep borehole into sediment deposited in front of power plant intakes. The project was a part of a bigger project titled Remediation of Dez dam reservoir owned by Khuzistan Water and Power Authority of Iran which was conducted from 2004 to 2008. The field measurements data available in KWPA (2004) are analyzed and discussed in this study.
Field Measurements
Field measurement was conducted to obtain the undisturbed and disturbed
samples from deposited sediment of Dez reservoir close to power plant intakes.
Sediments deposited in Dez reservoir near the dam body have always been present
as submerged sediments and thickness of water over the sediments has been more
than 50 m during the operating period of 45 years. In order to survey the physical
properties of sediment deposited in front of the power intakes, a deep borehole
was dug near power intakes located at 100 m upstream face of the dam. The digger
was installed on a barge, held by cables which were anchored to abutments as
shown in Fig. 3. A pistontype core sampler was used to obtain
samples. The sampler is operated by lowering it until the digger weight touches
the sediment surface. With the digger weight resting on the bottom, further
lowering of the sampler causes the digger arm to rise and release the coring
head. As the cutting sloe is just about to penetrate the sediment, the sampler
is penetrated into the sediment by hydraulic force, so, the sampler drives the
coring tube into deposited sediment. The piston remains fixed as the outside
tube moves past and serves to fold the undisturbed sample in the tube as it
is withdrawn. The depth of sampling is 63 m in maximum state. During the sampling,
the height of water above the sediment surface was 76 m (KWPA, 2004).
In the borehole, the collected sediments were very loose and disturbed down to the depth of 30 m. From depth of 30 to 63 m sediments were reported to be denser so undisturbed sampling was possible. This is a valuable experience for one is dealing with digging and differentiating the deposited fine sediments layers in large dam reservoirs.
In this study the bulk density of sediments was measured using ASTMD854 procedure and percentage of saturated water content was also determined by weighting method. The void ratio of the sediments was calculated based on the following relationship:
where, ρ_{sat} is the saturated density of the sediments, ρ_{w} density of water, G_{s} specific gravity of sediment solid. Based on the result of the experiments of the samples obtained from boreholes A and B, the average specific gravity of sediments (G_{s}) was 2.69.
The results of measured bulk density are presented in Table 2 and Fig. 4. The grain size distribution of obtained sediment samples from the boreholes is shown in Fig. 5.

Fig. 3: 
Photo of barge and digger located near Dez Hydropower intakes 
Table 2: 
Bulk density and void ratio of the borehole samples 

*Elevation of reservoir sediment surface in the time of sampling
was 256 m above sea level 

Fig. 4: 
Measured bulk dry density versus sampling level above sea
level 

Fig. 5: 
Grain size distribution curve for three samples 
RESULTS
Field measured data obtained from a deep reservoir are used to measured the
bulk density of deposited sediment and evaluate the available empirical methods.
To do this, the initial unit weight of deposited sediments was determined using
Eq. 1. Based on the grain size distribution curve of the borehole
samples (Fig. 5), the percentage of clay and silt content
of the samples were 55 and 45%, respectively, therefore:
W_{1} = 416x0.55+1120x0.45 = 732.8 kg m^{1} 
This value is an estimation of the initial bulk dry density for reservoir surface sediment layer that had a deposition time of less than 1 year. It should be noted that unit weight of sediment at the reservoir bed surface was measured a value of 825 kg m^{3} which has 11% deviation from the calculated value by the equation.
Real data of this study is used to evaluate the capability of empirical equations for prediction of the bulk density of deep sediment deposited behind Dez reservoir. The empirical methods of Lane and Miller were used for this purpose. The required time duration was calculated from the reservoir heightvolume curve which has been developed via reservoir bathymetry surveys in 1972, 1983, 1997 and 2003, as shown in Fig. 2. The time of deposited sediment can be estimated from the sediment level (Fig. 2). The deposition age of each layer, from its time of deposition to the year 2003 are presented in Table 3. Using these data, the bulk dry density of sediment samples was determined using equations of Lane and Miller in Table 3.
Table 4 shows capability of the empirical relations in prediction of the bulk density of deposited sediment. In Table 4, the measured bulk density has been compared with what it is predicted computed by Lane's method and Miller's method. The percent of error has been computed for each method. Figure 6 also shows computed bulk density versus the predicted one by Lane's method and Miller's method, respectively. As it can be seen from Table 4, both methods underestimate the bulk density of samples. The average errors for Lane's method and Miller's method are 43 and 46%, respectively.
Figure 4 shows increase of bulk density with sediment depth
in a growth rate of 0.75%. This is true because sediments in lower level have
been deposited earlier and have been compacted over time. Using Eq.
2 and 3, the bulk density grow rate with sediment depth
is surprisingly calculated to be 0.15 and 0.05% by Lane's method and Miller's
method, respectively.
Because of inaccuracy of the above empirical methods, the measured bulk dry
density versus elapsed time for the borehole is drawn in Fig.
6. Based on Fig. 6 it is possible to calibrate the Lane's
method and the Miller's method with the borehole data using the regression technique.
Table 3: 
Measured and calculated bulk dry density for samples from
the borehole 

Table 4: 
Comparison of measured and calculated bulk density (kg m^{3})
for samples from the borehole 


Fig. 6: 
The measured bulk dry density versus elapsed time for the
borehole 
The following relationships are obtained from the calibration of the Lane's
equation and Miller's equation, respectively.
W_{t}
= 739.74+460xLog (t) R^{2} = 0.948  (5) 
where, W_{t} is bulk dry density after t years of consolidation (kg m^{3}).
DISCUSSION
The constant values in the developed relationships (Eq. 5
and 6) refer to the initial bulk dry density of the sediments
(i.e., the bulk dry density of sediment surface) and their power coefficients
reference the B parameter in Table 1. In Table
5 the parameters are calculated from Eq. 5 and 6
and are compared with the values in Table 1.
Table 5: 
Comparison of the coefficients derived from calibration of
the empirical methods and the values obtained from Table 1 

^{a}kg m^{3}, ^{b}from Eq.
5 , ^{c}from Eq. 6 
The results show that the initial dry density calculated based on the Lane's
method has a low difference with the value obtained from Table
1, while the coefficient B shows a substantial difference. Therefore, the
error of the Lane's method and the Miller's method are significantly associated
with estimation of the B coefficient. From Table 5, it can
also be understood that the Lane's method is more accurate than the Miller's
method in practice. This is in agreement with the results in Table
4.
Most empirical methods are relayed on the sedimentation time variation and do not account for the effects due to depth variation of sediments and compaction of the sediments under the upper layers' loads. This is the main reason for inaccuracy of empirical methods particularly for reservoirs with high rate of deposition and thick layers of sedimentation like in Dez reservoir. To prove this hypothesis, the measured bulk densities reported by Bennett and Rhoton (2003) for Granada lake have been compared with the bulk densities calculated by empirical methods and the acceptable differences in range of 6 and 7.3% for Lane and Miller methods were obtained, respectively. The low thickness of sediments was the main reason for accurate estimation of empirical methods.
CONCLUSION
The conclusions of this study are summarized as follow:
• 
Sediment deposited near Dez dam body was fine sediment with
texture of 55% clay and 45% silt. The deposited sediments in upper layers
are very loose, disturbed and uncompacted in first 30 m 
• 
The bulk density of the sediments increases with depth almost
with the growth rate of 0.75%, while it is calculated to be 0.15 and 0.05%
by Lane's method and Miller's method, respectively 
• 
Comparison of the measured bulk density and calculated by
Lane's method and Miller's method show that both methods underestimate the
bulk density of fine sediments. The errors associated with Lane's method
and Miller's are estimated to be 43 and 46%, respectively 
• 
The calibration of the lane's equation and Miller's equation
show that the initial dry density calculated based on the Lane's method
has a low difference with the corresponding value obtained from Table
1, while they have substantial difference in prediction of the coefficient
B. The Lane's method shows to be more accurate than the Miller's method
for fine sediments 
• 
More investigation on processes of sedimentation in large
dam reservoirs is suggested in order to account for the depth variation
effects and development of an accurate empirical method for estimation of
dry bulk density and other physical properties of sediments 
ACKNOWLEDGMENTS
The authors acknowledge the Shahid Chamran University of Ahwaz and Shahrekord University of Iran and Khuzistan Water and Power Authority for financial support (Grant No.: KWPA82HM33) and facilitation of the experiments.