Professional Documents
Culture Documents
March, 2005
1. Introduction
1.1 Introduction
A model is a simplified representation of a real system, in terms of one or more mathematical equation. The most important word is simplified: a model can never represent the real system in all its details. Two main types of models exist. Empirical or statistical or black box models are based on experiments and statistical testing, not on theory behind the behaviour of the system under study. Deterministic models are based on (fixed) mathematical descriptions of natural processes governing the system behaviour. Field data are only used to validate the model performance or to calibrate certain model parameters. In general there are two types of questions you can answer with the help of a model: 1. why? questions: these focus on gaining insight in a water system, on cause-effect relationships, and on finding knowledge gaps; 2. what if?questions: these focus on the expected changes in the system as a result of measures or changes in the external conditions (e.g. climate changes). The use of a model is justified if such questions need a quantitative answer. In practice, the two types of questions often come together in one project. First the why question is answered to better understand the system, and afterwards relevant what if questions are addressed.
The present course does not have the scope of a hydro-informatics course. It concentrates on understanding and using the key equations in water quality modelling. It also provides an introduction into the use of modelling tools.
IHE
March, 2005
Environmental modelling
Step 1 (starting a logbook) is the starting point for obtaining a reproducible modelling study. Step 2 (defining the modelling project) is very important, because it involves the definition of the problem to be solved, as well as the definition of the objectives and the requirements for the modelling exercises. The underlying principle is that it is not possible to make a good model without knowing exactly which questions need to be answered and in what context. Relevant aspects can be: the type of problem, the time scales and spatial scales under consideration, the desired accuracy, and the constraints for acceptability of the model. Step 3 (building the model) consists of the selection of the conceptual model and of the actual construction of the model, either by programming the associated computer code yourself, or by using an existing (maybe commercially available) computer programme. The equations can be solved analytically (simple geometry, simple input data, exact solution) or numerically (all cases, approximation of the solution). The end of step 3 is the verification of the model: does the model solve the right equations in a correct way? With the model available, step 4 (analysing the model) includes activities like mass balance tests, sensitivity analysis, calibration, validation and uncertainty analysis. Not all of these activities are carried out in every study, but each one of them can be considered to improve a specific aspect of the models credibility. The word calibration refers to the adjustment of certain model parameters to make the model behave in agreement with field data. The word validation refers to the checking of the agreement between the model and the field data. Paying serious attention to step 4, helps you to carry out step 5 (using the model), which is where the actual purpose of the modelling activity starts. Step 6 (interpreting the results) provides the most vital input for the modelling study report, and ends with the following question: did the modelling study meet its objectives and requirements (defined during step 2)? The final step 7 (reporting and archiving) includes the formalisation of the end product of the study (the report) and forms the final step in the process of assuring the reproducibility of the work.
IHE
March, 2005
Qout, c
V, c
Qin, cin
Over a given period of time t, the water balance equation reads:
dV dt
t = Qin t Qout t
Dividing by t:
dV dt
= Qin Qout
dcV dt
March, 2005
Environmental modelling
dV c dt
dc + V = cinQin cQout dt
For the term dV/dt we can substitute Qin Qout, according to the water balance equation derived earlier:
dc c ( Qin Qout ) + V = cin Qin cQout dt dc V = cinQin cQout c ( Qin Qout ) dt dc V = Qin ( cin c ) dt
Dividing by V provides a first order differential equation for c:
dc Qin ( cin c ) = dt V
dc = kc dt
With k the decay rate (1 per time, s-1). When we include a decay term, the pollutant balance equation is written as follows:
dc Qin ( cin c ) kc = dt V
IHE
March, 2005
Qin dc c kc = Ec = V dt
Where E is a constant number. The solution to this equation is:
c = ct =0 exp ( Et )
For more complex cases analytical equations may be obtained from the literature, but the use of numerical methods provides an alternative.
dc Qin ( cin c ) = dt V
The core of the Euler method is the approximation of the time derivative:
dc ci +1 ci t dt
If we substitute this in the pollutant balance equation we obtain:
ci +1 ci Qin ( cin c ) t V i
Now we solve for the concentration at the end of the interval t:
Q ci +1 = ci + t in ( cin c ) V i
If we provide the initial concentration c(t=0), the formula above can be used to evaluate the concentration as a function of time, by proceeding with time steps t.
IHE
March, 2005
Environmental modelling
3. BOD and DO
3.1 Sources and sinks of DO
The principal DO problems stem from municipal and industrial BOD discharges, the oxidizable nitrogen forms, and nutrients which may stimulate phytoplankton growth. The nature of the aquatic ecosystem then determines the DO levels through such processes as reaeration, photosynthesis and sediment oxygen demand. The DO problem begins with the input of oxygen demanding wastes into a water body. In the water body itself, the sources of DO are: 1. Reaeration from the atmosphere 2. Photosynthetic oxygen productions 3. DO in incoming tributaries or effluents. Internal sinks of DO are: 1. 2. 3. 4. Oxidation of carbonaceous waste material Oxidation of nitrogenous waste material Oxygen demand of sediments of the water body Use of oxygen for respiration by aquatic plants and algae.
With the above inputs and sources and sinks, the following general mass balance equation for the DO concentration (designated by c) in a water volume V, can be written as follows:
dc = dt
reaeration + (photosynthesis - respiration) - oxidation of CBOD, NBOD (from inputs) - sediment oxygen demand + oxygen inputs + oxygen transport (into and out of segment) (2.1)
The various source-sink components will now be examined in order to further develop the mass balance equation.
Theoretically, 6x32/180 = 1.07 mg O2/mg glucose are necessary. In practice, the CBOD is only a fraction (40-80%) of this theoretical value, due to e.g. the partial non-biodegradability of the material. Also, in the laboratory procedure the BOD degradation may be measured only for the first 5 days; in this way a BOD520 , depicting the BOD after 5 days at 20 C, may be defined.
6 IHE
March, 2005
Nitrogeneous materials (e.g. proteins, urea, amino acids) are first hydrolyzed to ammonia, NH4+. Via the nitrification reaction this is then oxidized to NO3-: NH4+ + 2 O2 NO3- + 2 H+ + H2O Theoretically, 4.57 g O2 are used per gram NH4+-N. In practice, some of the ammonia will be used for cell production, and the above oxygen utilization will have a value of 4.2-4.5. Kinetics of BOD degradation Although CBOD and NBOD may have different degradation kinetics, they are, as a first approximation, often combined. Assuming also first order decay, then: dL/dt = - k1 L , in which: L = remaining BOD at time t (mg/L) k1 = BOD decay rate constant (day-1) Integrating; at t = 0, L = L0: L = L0 exp (-k1 t) (2.2)
K1 = 0.1 - 0.4 day-1 (dependent on e.g. river characteristics, wastewater purification, time after discharge). The temperature effect on k1 is often expressed as: k1(T) = k1(20) 1.04(T-20)
March, 2005
Environmental modelling
dc/dt = k2 (cs c) k2 (day-1) is the reaeration coefficient; c = cs - (cs c0 ) exp (-k2 t) integrating: (2.3)
The reaeration coefficient k2 is a function of: Turbulence caused by currents and waves Waterfalls, dams and other hydraulic structures Surface films, e.g. by detergents Temperature, often expressed as: k2(T) = k2 (20) 1.02(T-20)
In many rivers over the world, the k2 behaviour has been studied in detail. The following semiempirical formula is commonly used to describe k2 as a function of water depth H and water velocity U:
bUm k2 = Hn
(2.4) in which b, m and n are constants. It can tentatively be shown that the power of H will be >1: reduced water depth gives rise to higher water turbulence; also, a proportionally smaller water volume has to be supplied with the oxygen. The power of U will be between 0.5 and 1. In other words, reaeration is much more important for shallow, fast flowing rivers. Over the years the results of three main researches have been used (Thomann & Mller, 1987, cf. Fig 2.2): O'Connor (1958) for relatively deep, moderate velocity streams: b = 3.863 , m = 0.5 , n = 1.5 (for U in m/s and H in m) Churchill (1962), for deep, fast flowing rivers: b = 5.026 , m = 0.969 , n = 1.673 (for U in m/s and H in m)
IHE
March, 2005
Owens (1964), for shallow streams: b = 5.322 , m = 0.67 , n = 1.85 (for U in m/s and H in m).
Fig. 2.2. Reaeration coefficient k2 (day-1) at 200C as a function of water depth (feet) and velocity (feet/s) (1 foot = 0.3 m). Data were taken from three main researchers, using Eq. 2.4.
IHE
March, 2005
Environmental modelling
10
IHE
March, 2005
We assume that the discharge Q in the river is constant and homogeneous. Therefore, the inflow is equal to the outflow for all river segments: Qin = Qout = Q.In a river, the discharge can be expressed as the cross section A (m2) multiplied with the velocity U (m/s, average over the cross-section):
Q = A U
The volume of each of the river segments is equal to the cross-section A multiplied with the segment length x:
V = A x
Therefore, we can write U/x rather than Q/V in the mass balance equations for the river segments. We apply the mass balance equations to the individual segments, in a downstream direction. For BOD:
U U U Lin L1 L2 x x x , L2 = , L3 = , etc. L1 = U U U + k1 + k1 + k1 x x x
and for DO:
U U U cin k1 L1 + k2 cs c1 k1 L2 + k2 cs c2 k1 L3 + k2 cs , c2 = x , c3 = x , etc. c1 = x U U U + k2 + k2 + k2 x x x
IHE
11
March, 2005
Environmental modelling
Note: these equations can be solved analytically as well, by the famous Streeter Phelps equations:
12
IHE
March, 2005
dColi = kColiColi dt
where the kColi increases with: Intensity of UV radiation Increase of Water temperature Increase of Salinity
The minimum value is in the range of 0.1 d-1. UV radiation is a very important factor. In clear water, for instance sea water, and at high radiation intensity, mortality rates up to and over 50 d-1 have been observed. It is important to realise that the UV-radiation at the water surface can not penetrate fully into the water column. In practice, the intensity of the radiation varies with the depth, as follows:
I ( z ) / I 0 = exp( ez )
where: I I0 z e intensity of radiation (W/m2) intensity of radiation at water surface (W/m2) distance from water surface (m) extinction coefficient for UV light (m-1)
The value of e depends on the turbidity of the water. Typical values range from 0.2 (very clear) to 5 (very turbid). A mathematical expression for the coliform decay rate can be obtained from Mancini, J. L., 1978. Numerical estimates of coliform mortality rates under various conditions. Jour. Water Poll. Control Fed. : 2477-2484.
IHE
13
March, 2005
Environmental modelling
vs dSS = H dt
where: SS vs H cr
1 SS cr
concentration of suspended particles (g/m3) settling velocity (m/d) water depth (m) shear stress (N/m2), exercised by currents, function of velocity critical shear stress (N/m2)
This formula represents a decay caused by settling of particles to the bottom. Note that the equivalent decay rate vs/H is inverse proportional to the water depth. Values of the settling velocity are in the range of 0.1 to 10 m/d. The shear stress can be approximated by:
C1V
where V is the velocity in m/s, and C1 is in the range 1-10 (e.g. 4). The factor (1-/cr) expresses that strong currents cause a lot of turbulent mixing which keep the particles in suspension, while weak or absent currents allow particles to settle. The critical shear stress depends on the grain size of the sediments and the degree of consolidation. It is in the range 0.2-5 N/m2.
14
IHE
March, 2005
sedimentation
Formulating a model including these processes reaches too far in this course. By approximation, nitrogen and phosphorus can be considered as decaying pollutants, where the decay consists of: denitrification at the water sediment interface (nitrogen); settling of particles with a significant phosphorus content.
vN dN = N H dt
vP dP = P H dt
The parameters vN and vP are strongly dependent on the water system in question, and need to be derived from field data.
IHE
15
March, 2005
Environmental modelling
P=
Cdiss Csolid P
C diss C solid
concentration of metals, dissolved (g/m3) concentration of metals, solid phase (g/g of suspended particles) partition coefficient (m3/g)
The heavy metal fraction in the particles can settle to the bottom, together with the particles. The mathematical expression for the heavy metal (HM) reads:
vs dHM dSS = P = P 1 SS dt dt H cr
16
IHE