I. INTRODUCTION
For efficient controlling
of COD in industrial waster water treatment plants it is necessary to get rough
information about the time distribution of incoming waste-water and incoming COD
charge. In contrast to municipal treatment plants waste-water in industrial
areas is discharged with a highly dynamic time distribution due to the dynamic
operation of plants and waste-water discharge in the production processes operated
by the business units in the area. Incoming COD charge depends mainly on the
time distribution and mass of discharged waste-water from every unit in the industrial
area, on the network topology, and on the size of internal reservoirs in the
network. To avoid exceeding COD limits in the inlet of the waste-water
treatment plant it is a good idea to monitor COD at selected points of the
network and balancing COD peaks with internally available reservoirs. Model
based computation of flow and mixing processes in the network should be added
to achieve a highly efficient plant operation. The simulation method discussed
here is comparatively simple and can be established with low effort.
II. METHOD
To achieve an efficient
operation of the treatment plant, the waste-water network is divided into
discharger, pipes, reservoirs, nodes and control elements (pumps, valves, etc.).
The classical approach is to set up the conservation equations for impulse,
overall mass, component mass and energy in every point of the network and to solve
the resulting equation system with numerical methods. This procedure results in
mass flow, component concentration and temperature inside the network (see for
example [1-3]). Several modern plant simulation tools are available worldwide,
mainly for simulation of stationary operation points. Simulation of transient
operation is difficult due to complexity of physical parameters, discrete types
of flow events, numerical problems and excessive large computation times.
In industrial waste-water
networks the discrete type of flow events is a main problem for classical
simulation. Due to discharging from discontinuous operated processes a large
amount of the network topology is empty most of the time and flow events occurring
like packages of waste, which are send through the network. The modelling
method mentioned here pick up this type of event and try to follow the ‘waste
packages’ through the network from node to node up to the next reservoir.
III. MODEL SETUP
The system consists of
-
waste-water
pipes,
-
nodes and ducts,
-
reservoirs
(sampling and clearing pools),
-
control
equipment (pumps, valves),
-
discharger
sources (production units).
The mathematical model
should describe
-
the volume flow
and
-
the COD
concentration (or the COD load)
in
the considered branches of the network as a function of time. Additional it is
possible to achieve the temperature distribution in the network as a function
of time by implementing energy balances. By doing this one should be clear,
that also mostly COD-free discharges like washing water must be taken into
consideration to get correct temperature information. This effort must be
checked against the advantages of having knowledge about temperature dependencies
in the network flow.
A static time grid is
chosen as a discrete scheme for numerical stability reasons. Due to the
numerical robustness this method has high advantages over classic numerical integration
methods via time step control. The following Figures 1 to 3 give an impression
of the approach.
First (see Fig. 1) the discharge
of a specific waste-water volume (grey element) from a source in a pipe at
time 
is illustrated. The
discharged waste-water ‘package’ is characterized by its volume, its COD
concentration and its temperature.
Figure 1. Discharge
of a waste-water ‘package’ at the time
In the following time step 
the considered volume
element has travelled a distance through the pipe given by its flow velocity;
simultaneously a second volume element is discharged in the pipe from the same
source (see Fig. 2).
Figure 2. Discharge
and flow at the time
The volume element considered
in Fig. 1 gets to the end of the pipe at time 
and is added to the
reservoir (see Fig. 3). Simultaneously other volume elements are discharged
from the source and travelling through the pipe.
Figure 3. Element
discharged at 
arriving the next
reservoir at the time 
It’s an important
assumption of our approach, that every pipe in the network has a characteristic
constant hold-up time 
for all volume elements. Waste-water is sent in discrete
packages through the pipe and other physical effects e.g. diffusion or back mixing
are not taken into consideration. Depending on the network topology the
reservoir illustrated in Fig. 3 can also be a pipe node (this is a reservoir
with zero volume). In Figures 1 to 3 also a volume element discharged and travelling
from the reservoir is illustrated. This happens simultaneously with the discharge
from the upstream source, and the reservoir is treated as a discharge source as
mentioned above with a specific COD and temperature given by simple conservation
requirements.
The reflection given above
for a single discharger/pipe/reservoir element is done simultaneously for all
elements in the network topology, which is only easy computer work also for bigger
networks. A second important simplification can be the assumption, that the
hold-up time is equal for all pipes
and equal to the computing time step 
. This means, that all volume elements discharged from a
source, reservoir or node are travelling to the next node exactly within a
single computation step . This simplification makes the simultaneous computing of
elements more easy, but it is not essential for the approach and must be
verified for specific cases.
For the given model
elements pipe, node, reservoir and pump the following equations results from
conservation requirements:
Pipe
Node
Reservoir
Pumps
As mentioned above, in many
cases the computing step can be set equal to the pipe hold-up time:
The mean pipe hold-up time depends
from the volume flow and the pipe topology and can be considered from channel hydraulic
calculations (see [3]). Example given: for a channel with diameter 300 mm and a
hydraulic incline of 0.8 % at volume flows above 5 m3/h the
hold-up time will be in the range of 2-3 minutes for a pipe length of 100 m.
IV.
CHARACTERIZATION OF DISCHARGE
In most cases gauging
equipment for recording waster water volume and quality will be not available
for every industrial production company located in an industrial area. Anyway
information for simulation of the network can be achieved by collecting information
about the production processes and their waste-water fluxes, and combining them
in specific periodical event diagrams for every source. It is useful to
distinguish between
-
base load
discharges, lasting for longer times, and
-
peak load
discharge events, lasting only a limited amount of time.
Peak loads set up on the
base load and establish a family of periodical functions, from which a sequence
of events can be constructed. Fig. 4 illustrates how to describe the time dependent
discharge function of a source by summing up a base load and two periodical
peak load events.
