Which "Tank Mixing Model" attribute should I use? How do each of them work?
First, keep in mind that all tank mixing models are approximations.
Most standard elevated tanks are close to completely mixed unless they are very large.
Tall skinny metal standpipes are the tanks most likely to stratify. In that case the LIFO (Last in first out) or 2 compartment models are best.
Tanks with separate inlet/outlet pipe depend on the configuration of the tank. If there are internal baffles to maximize detention time such as chorine contact tanks, you may approach the FIFO (First in First out) model. If there is plenty of inlet momentum and the tank is small, then the complete mix model may be best. If there are dead zones, you may need the 2 compartment model.
For a top-fill tank, FIFO typically would make sense because the parcels of water that first enter at the top have to drain down vertically before they leave the bottom. Consider running simulations to test out both the FIFO and 'fully mixed' options to see how the results compare. If they are similar then you don't need to worry which option you choose. If you still have some concerns then you can consider the next steps (i.e. more detailed CFD analysis of the tank, installation of a mixing system, installation of a chlorine booster, etc).
Note that there are two very different types of tanks and they have different goals: 1. Tanks before the first customer are intended to provide contact time to meet CT requirements. You want to maximize average contact time and this is done using baffles to approach plug flow. You would model this with a FIFO approach with a separate inlet and outlet. 2. Tanks in the distribution system can cause a deterioration in water quality. You want to minimize average detention time and this is done by trying to achieve complete mixing. You can get good mixing by maximizing inflow velocity with a jet inflow in the bottom and the same pipe for inlet and outlet. Avoid very tall skinny tanks as these can stratify in warm weather. You need to have a lot of turnover. A good report on this is Grayman W. et al. (2000) "Water Quality Modeling of Distribution System Storage Facilities,' AWWARF. It give some good guidance on the amount of turnover you need to get complete mixing. There are a number of manufacturers of in-tank mixers if you can't get sufficient mixing from the inlet jet.
Real water distribution tanks cannot be exactly described as plug flow or completely mixed but these are reasonable approximations to fluid behavior in tanks. WaterGEMS and WaterCAD support four types of tank mixing models. The tank attribute Tank Mixing Model allows you to select from the following options.
The Completely Mixed model assumes that all water that enters a tank is instantaneously and completely mixed with the water already in the tank. It applies well to a large number of facilities that operate in fill-and-draw fashion with the exception of tall standpipes.
The 2-Compartment Mixing model divides the available storage volume in a tank into two compartments, both of which are assumed completely mixed. The inlet/outlet pipes of the tank are assumed to be located in the first compartment. New water that enters the tank mixes with the water in the first compartment. If this compartment is full, then it sends its overflow to the second compartment where it completely mixes with the water already stored there. When water leaves the tank, it exits from the first compartment, which if full, receives an equivalent amount of water from the second compartment to make up the difference. The first compartment is capable of simulating short-circuiting between inflow and outflow while the second compartment can represent dead zones. The user must supply a single parameter, which is the fraction of the total tank volume devoted to the first compartment. This value can be determined during calibration if this model is selected.
The FIFO (First In, First Out) Plug Flow model assumes that there is no mixing of water at all during its residence time in a tank. Water parcels move through the tank in a segregated fashion where the first parcel to enter is also the first to leave. Physically speaking, this model is most appropriate for baffled tanks that operate with simultaneous inflow and outflow such as ideal clear wells at water treatment plants. There are no additional parameters needed to describe this mixing model.
The LIFO (Last In, First Out) P lug Flow model also assumes that there is no mixing between parcels of water that enter a tank. However in contrast to FIFO Plug Flow, the water parcels stack up one on top of another, where water enters and leaves the tank on the bottom. This type of model might apply to a tall, narrow standpipe with an inlet/outlet pipe at the bottom and a low momentum inflow. It requires no additional parameters be provided. The calculated age reported at a tank using the LIFO mixing method is the age at the inlet of the tank.
Book: Advanced Water Distribution Modeling and Management page 53.
Wiki: Does the Top Fill Tank property factor into mixing?
Wiki: How does WaterCAD/GEMS determine the age of the tanks using the LIFO or 2-compartment method?
Forum discussion: Water Age and Tank Mixing - Evaluating the Benefit of a Mixing System