CASE STUDIES AND VAlUE OF RESEARCH
Application of Storage Facility Modeling to Improve
Mixing and Reduce Water Age
David J. Hartman and Jeff Swertfeger, Greater Cincinnati Water Works, ohio
Sometimes the solution to DBP problems lies not only in adjusting treatment and water chemistry, but also
reducing water age in the distribution system. Water Quality Modeling of Distribution System Storage
Facilities (2000, order#90774/project#260) describes the influence of water storage facilities on water age. This
report describes how it is not just a simple calculation of average residence time in a facility that is important,
but that the mixing characteristics inside the facility can also greatly influence the age of water leaving the tank.
Improper mixing can cause high water age, which can lead to low chlorine and high DBPs in the tank effluent.
This project is important not only because it describes these problems, but because it also contains two software
tools (Hydro Tank and Comp Tank) to assist utility staff in assessing mixing in their storage facilities and in designing
solutions to mixing problems.
The Greater Cincinnati Water Works (GCWW) has continuous chlorine monitors at the effluent of most of their
tanks. These monitors are connected to a supervisory control and data acquisition (SCADA) system, which tracks
the data and GCWW sets goals and action levels based on these monitors. This monitoring and evaluation
system is part of an overall strategy to control water age, thus maintaining the state-mandated chlorine level and
In evaluating this data, GCWW noticed steep drops in chlorine levels at one particular tank that occurred when the
level in that tank dropped to a certain elevation. This drop in chlorine was much greater than would be expected
based on calculated water age using simple residence time calculations. This led GCWW to believe that stratification
may be occurring in the tank, a zone of stagnant water was forming under normal operation, and the water from
this zone would leave the tank only when the tank level dropped below a certain elevation. To confirm this, GCWW
turned to the Hydro Tank model included in Major Ion Toxicity in Membrane Concentrates (2000, order #90824/
project #290) and developed a simple computational fluid dynamics (CFD) model.
The tank itself is a 2-million gallon tank with a 100-foot diameter and a maximum water depth and overflow 35
feet from the bottom. The main leading to the tank has a diameter of 2 feet, but the common inlet/outlet riser
of the tank had a diameter of 5 feet. Upon running the model, GCWW was able to see that the velocity of water
going into the tank was inadequate in a typical fill and drain cycle to break the stratification largely because of
the large diameter of the inlet. The model demonstrated that the influent water was contained in an area directly
above the inlet and very little if any mixing occurred with the rest of the water in the tank. In a drawdown cycle,
this is the first water to leave the tank, i.e., the last water in is the first water out. In extended drawdowns, the
older water from the upper zones would also leave the tank resulting in low chlorines measured in the tank
effluent. For a 5-foot diameter inlet, the model demonstrated a fill time of over 20 hours would be required to
completely mix the tank. However, the tank is typically operated with a fill time of around 8 hours. This 20 hour
fill time would also exceed the capacity of the tank during a normal tank cycle. This led GCWW to realize that
operationally there was little it could do to mix the tank to address water quality issues and still meet operational
pressure and flow requirements.
As a result of the modeling, GCWW realized that a modification to the tank’s inlet diameter was needed to allow
mixing. GCWW used the model to experiment with several different configurations to balance the need for mixing,
but still maintained pressure and flow requirements over the entire range of operation. GCWW also needed to
be sure that restricting the influent diameter would not create significant enough headloss that would affect