DIMENSIONLESS STEADY-STATE NSOD MODEL

Carlo Gualtieri
Hydraulic and Environmental Engineering Department “Girolamo Ippolito”
University of Napoli“Federico II”. Via Claudio, 21 80125 Napoli (Italy)
Phone: 0039-081-7683460 Fax: 0039-081-5938936 E-mail: cagualti@unina.it

Abstract: Oxygen demand exerted by bed sediments of rivers, lakes and estuaries (SOD), which is due to
the oxidation of settled organic matter, could often greatly affects their water quality. Thus, a proper
estimation of SOD amount is a relevant goal in dissolved oxygen balance modeling. However, SOD
available modeling frameworks remain far from a complete understanding of complex phenomena
involved.
The paper proposes a dimensionless steady-state model of nitrogenous sediment oxygen demand
(NSOD), where the solutions of mass balance equations for organic carbon and ammonium provide an
useful estimation of system response. Moreover, a careful model parameterization allows to introduce
some simplifications in the solution obtained.
Keywords: environmental hydraulics, mass-transfer at water-sediment interface, diffusion, sediment
oxygen demand, dimensionless model

1 FOREWORD
Sediment oxygen demand (SOD) is a key process in dissolved oxygen balance in river, streams and lakes
(Thomann and Mueller, 1987; Chapra, 1997; Gualtieri, 1998). This process is due to the decomposition of
organic matter in bottom sediments of rivers, lakes and estuaries, which produces CH 4, CO2 and H2S to
being subsequently oxidized, resulting in a consumption of dissolved oxygen formerly present in the
overlying water column; these benthic deposits could derive both from point and non-point, i.e.
distributed sources, which contain high concentrations of organic particulates. Among the former, the
most remarkable are raw sewage or combined sewer overflows (CSO), while among the latter, should be
considered allochthonous particulates such as leaf litter, plants roots, eroded organic-rich soils and, in
highly productive environments as eutrophic water bodies, photosynthetically produced plant matter.
Moreover, ammonification process transforms organic carbon into ammonium that is subjected to
nitrification with consequent oxygen demand from water volumn. Also, some benthic organisms other
than common bacteria, such as the attached, filamentous organism called Sphaerolitus or the freshwater
mollusks called zebra mussels, could be responsable of a remarkable SOD.
In shallow and speed streams with sandy and gravelly bed, which have low affinity for sorption, fine
organic particulates tend to remain suspended, thus organic sediments fraction is low and sediment
oxygen demand is negligible. On the contrary, if bed shear stress drops below a threshold value, which
could be related to a flow velocity of 0.30 m/s, fine organic particulates and other finer particles settle to
the bottom, increasing the organic content of bed sediments. In some rivers and estuaries, settling
phenomena could occur only during summer and autumn with low water flowrate, while for high
flowrate the bed could be partially resuspended.
However, when settling phenomena are present, the oxygen demand exerted by bed sediments could be
of considerable importance and could greatly affect the dissolved oxygen levels in the water column.
Nevertheless, SOD available modeling frameworks remain far from a complete understanding of complex
chemical, biological and physical phenomena involved and so are unable to provide an adequate
quantitative assessment of resulting oxygen demand exerted; particularly, while in water quality models
it is possible to describe and to quantify particulate organic matter flux from water column to bed
sediment by settling, so that we can predict how much organic particulate reaches the sediments, the
mechanism whereby the sediment organic is converted into oxygen demand remains not clearly defined
(Chapra, 1997).
The paper proposes a dimensionless steady-state model for NSOD, that is based on mass balance
equations for organic carbon and ammonium. The model provides an estimation of the fraction of
organic particulate settled on the bed that results in organic carbon and ammonium concentrations
within the sediments.

2 SEDIMENT OXYGEN DEMAND. STATE-OF-THE-ART

Several models have been proposed to explain SOD production and to obtain a prediction of its value;
firstly, some empirical relationships between SOD and sediments volatile solids concentration or
chemical oxygen demand (COD) were attained, resulting in a square root dependency (Fair et al., 1941;
Gardiner et al., 1984). Then, this dependency was confirmed by some analytical models that exhibit
some shortcomings in a proper parameterization (Bouldin, 1968; Walker and Snodgrass, 1986).
Finally, diagenesis model has provided an analytical modeling framework that is able to explain in
mechanistic fashion the observed square-root relationship of SOD to sediment organic carbon content
(Di Toro et al., 1990). Key point is the flux of particulate organic matter to the sediments and its
decomposition or diagenesis which produces a flux to being modeled of some reduced species, such as
dissolved methane, ammonium and H2S. The framework starts from the assumption that bed sediment
could be divided into two layers, where the reactions resulting ultimately in SOD occur; the former is
aerobic and oxidation processes take place, while the latter, underlain the former, is anaerobic, with
reduction reactions. The considered reactions concern with particulate carbon organic and particulate
nitrogen organic settled to sediments. In the anaerobic layer, organic carbon is decomposed to yield
dissolved methane CH4, which diffuses upward to the aerobic zone where it is oxidized, resulting in a
carbonaceous sediment oxygen demand (CSOD). In the same way, in the anaerobic layer, ammonification
reaction of organic nitrogen produces ammonium that is, then, nitrified in the aerobic layer to result in a
nitrogenous sediment oxygen demand (NSOD), while ammonium that is not converted to nitrate diffuses
in the water column; however, part of nitrate is denitrified to form nitrogen gas using as carbon source
the methane, decreasing the final CSOD. Particularly, if dissolved methane concentration in the
anaerobic layer never exceeds saturation, the relationship between CSOD and organic carbon flux J C is
linear; otherwise, if saturation value is exceeded, the analytical solution of model, if methane is
completely oxidized in the aerobic zone, is:

(1)
where csat is the saturation concentration of dissolved methane, J C* is the organic carbon flux from water
column by settling, in oxygen equivalents, and D=DCH4/H is a mass-transfer coefficient for dissolved
methane, where DCH4 is the diffusion coefficient, [L² T-¹], and H is the thickness of the active sediment
layer. Eq. (1) holds if dissolved methane is completely oxidized, but in reality in the aerobic layer a
competition exists between oxidation reaction and diffusion transport, thus, if the former is much slower
than the latter, much of methane would merely pass through the aerobic zone, resulting in a CSOD
decrease respect to (1). Thus, final CSOD equation is:

(2)
where O2(0) is the oxygen concentration at the water-sediment interface and C is:

(3)
where kCH4 is the decomposition rate of dissolved methane in the aerobic layer, [T -¹], and DO2 is oxygen
diffusion coefficient, [L² T-¹]. This coefficient is approximately 2.1×10-9 m²/s. In effect, (1-sech) term
accounts for the outlined competition as well as the shut down of CSOD when oxygen in the overlying
water column gets low.
Using a similar analysis, Di Toro model also evaluates the effect of nitrification on SOD; in this analysis
NSOD is modeled starting from carbon flux and, thus, final expression of SOD is:

(4)
where both CSOD and NSOD are considered and N is:

(5)
where kN is oxidation rate of ammonium to nitrogen gas, [T -¹], and DN is ammonium diffusion coefficient
in water, [L² T-¹]. Moreover, in (4), ron=1.714 g O/g N is the oxygen demand for nitrification, corrected for
denitrification, and ano=0.0654 g N/g O is the nitrogen yield from the decomposition of settled organic
matter.
Inspection of (4) shows that the conversion of organic carbon flux into nitrogenous oxygen demand, i.e.
ron×ano×JC*, has not a square-root dependency from J C* because of lack of gas formation in the anaerobic
layer; then, the hyperbolic secant term in NSOD expression has the same meaning as for CSOD.

3 NSOD MODEL PRESENTATION

Diagenesis processes can be modeled also through mass balance equations for organic carbon and for
ammonium; then, these equations could be solved applying numerical solution techniques, which offer
greater flexibility and broader application than analytical methods.
This paper deals with only ammonium diagenesis that could be modeled representing the active
sediments as two well-mixed layers. Also, the flux of particulate organic matter delivered to the
sediments by settling is represented in carbon units, as a flux of particulate organic matter, rather in
oxygen equivalents as in the analytical model previously outlined. In fact, organic matter could be
represented as formed, among other elements, by carbon and nitrogen that are present in a ratio
dictated by stoichiometry. Thus, when organic matter is decomposed, decomposition of organic carbon
results in a fixed yield of organic nitrogen. In particular, a nc=0.176 g N/g C, is the stoichiometric yield of
nitrogen from the decomposition of a gram of organic carbon.
In the model here proposed the downward flux of organic matter is in carbon units and the
ammonification reaction of organic nitrogen is expresses using organic carbon diagenesis process.
Sediments volume where diagenesis occurs is formed by two layers. The upper layer is the aerobic layer,
while the lower layer is the anaerobic one (Fig.1). Then, mass balance equations could be written for
organic carbon flux from column water to sediment and for ammonium in the anaerobic and aerobic
zones. If Vaer and Vanaer are, respectively, aerobic layer and anaerobic layer volumes, [L³], A is the area of
water-sediment interface, [L²], JC is organic carbon flux from water column by settling, [M L-² T-¹], kc-anaer is
the first-order diagenesis rate of organic carbon, [T-¹], and canaer is the concentration of organic carbon, [M L-
³], mass balance for organic carbon in the anaerobic layer is (Chapra, 1997):

(6a)

Fig.1 Numerical NSOD model

Dividing both sides by A, (6a) yields, if Vanaer=A Hanaer :

(7a)
where Hanaer, in m, is the thickness of the anaerobic layer.
Similar flux balances can be written for ammonium in the aerobic and anaerobic layer, respectively, as:

(7b)
 (7c)
where naer and nanaer are the ammonium concentration in the aerobic and anaerobic layers, respectively,
[M L-³], kn-aer is the nitrification rate of ammonium in surface layer, [T -¹], Dn is the diffusion coefficient for
ammonium in water, that is equal to 9.80×10 -10 m²/s, while Haer, in m, is the thickness of the aerobic layer.
Finally, Haer-anaer and Hwater-aer, in m, are mixing lengths for diffusive transfer process. Notably, in (7c) the
ammonium concentration in the overlying water is assumed to be negligible.
Moreover, as organic carbon enters the system as particulate matter, the incoming flux is J C=cin×vs, where
cin is the concentration of particulate organic carbon in the water, in g/m³, and v s is its settling velocity, [L
T-1]. Generally, vs lies in the range from 0.1 to 0.5 m/day, that is 1.16÷ 5.88×10 -6 m/s, while cin could vary
in a wider range, from 0.5 to 100 g/m³ (Chapra, 1997). At steady-state, (7a), (7b) and (7c) yield:

(8a)

(8b)

(8c)

4 DIMENSIONLESS MODEL. MODEL SOLUTIONS

System of equations (8a), (8b) and (8c) could be rewritten in dimensionless fashion in order to point out
the mechanisms that finally transform particulate organic carbon into NSOD. In fact, dimensionless
solutions provide, instead of organic carbon and ammonium concentrations, their ratio to particulate
matter input concentration cin. These ratios are transfer functions bc= canaer/cin, bn-aer= naer/cin and bn-anaer=
nanaer/cin (Gualtieri and Pulci Doria, 1998a, 1998b). The transfer functions bc, bn-aer and bn-anaer could be
obtained dividing by vs×cin. Then, (8a), (8b) and (8c) can be rewritten in dimensionless fashion as:

(9a)

(9b)

(9c)
Notably, for water column-aerobic sediments layer pore water exchange, i.e. diffusion, the characteristic
mixing length could be taken, as usual, to be the depth of surficial benthic segment, while for aerobic
layer-anaerobic layer the mixing length is the distance between the midpoints of those layers. Therefore,
Hwater-aer=Haer and Haer-anaer=(Haer+Hanaer)/2. Also, these equations comprise 6 input parameters to be
estimated: vs, kc-anaer, kn-aer, Dn, Hanaer, and Haer.
These dimensionless equations could be solved for:
 (10a)

(10b)

(10c)
where only 3 dimensionless parameters are present.
Inspection of (10b) and (10c) shows that some simplifications could be introduced. In fact, recalling (5),
since Dn=9.80×10-10 m²/s, DO2=2.1×10-9 m²/s and kN»1×10-5 s-1, it results that kn-aer is of the order of 10-² s-1.
Notably, nitrification rate in the aerobic layer is higher than generally occurs in the water, where k n-aer lies
in the range from 1.0×10-7 to 5.0×10-5 s-1. This is not surprising since sediments have much higher
bacterial biomass than the water. Moreover, as the thickness of aerobic layer H aer is of the order of 10-³
m, (kn-aer×Haer²)/Dn»10². Thus, it results that (1+(kn-aer×Haer²)/Dn)»(kn-aer×Haer²)/Dn. Therefore, (10b) and (10c)
yield:

(11b)

(11c)
Moreover, in (11c), (Haer+Hanaer)/2>>Dn/(kn-aer×Haer), resulting in a further simplification:

(12b)
Interestingly, those simplified solutions, i.e. (11b) and (12c), could be directly derived from (9a), (9b) and
(9c). In fact, assuming that nanaer>>naer, equation (12b) holds. Also, if ammonium flux from aerobic layer to
water column is negligible, then equation (11c) could be derived.
Finally, since Hanaer>>Haer, (Haer+Hanaer)/2»Hanaer/2 and (12b) yields:

(13b)
Therefore, model solutions are:

(10a)

(13b)
 (11c)
Careful inspection of (10a), (13b) and (11c) shows that transfer function for organic carbon in the
anaerobic layer and for ammonium in the aerobic layer have a similar fashion, where an index number
vs/(kc/n-anaer/aer×Hanaer/aer) is present. This number relates organic carbon flux to decay first order constant
kc/n-anaer/aer and thickness layer Hanaer/aer. Also, for ammonium in the anaerobic layer, transfer function b n-anaer
increases with settling rate vs and thickness layer Hanaer and is reduced by diffusion process.

5 APPLICATION OF MODEL. RESULTS

The equations derived from the foregoing analysis could be applied as prediction tool for NSOD. In fact,
they provide final value of organic carbon and ammonium in the considered layers. However, among the
dimensionless solutions obtained, (11c) is the most relevant in NSOD estimation, because this equation
relates the concentration of particulate organic carbon with the resulting ammonium concentration in
the aerobic layer that ultimately produces NSOD. Fig.2 shows a plot with b n-aer×kn-aer vs Haer for different
values of settling velocity vs according (11c). Value of settling rate are in literature range. Model solutions
are all comprised in a narrow band.

Fig.2 Transfer function bn-aer

6 CONCLUSIONS
Sediment oxygen demand (SOD) is an active and rapidly evolving research area of environmental
hydraulics. Available modeling frameworks are yet unable to provide a comprehensive insight in the
complex processes involved in SOD production.
The paper has proposed a dimensionless steady-state model addressed to explore key parameters
involved in ammonium nitrification, that represents an oxygen-demanding reaction that exerts a
nitrogenous sediment oxygen demand (NSOD) and could be a relevant quantitative influence on overall
SOD formation. Dimensionless solutions has been obtained solving mass balance equations. These
solutions have pointed out how incoming particulate matter is transformed into organic carbon and
ammonium concentrations within the sediments. Particularly, model solution for ammonium in the
aerobic layer could be applied as predictive tool for NSOD. Finally, the paper has provided simplified
solutions that offer a faster parameterization of the model.