WaterCAD - water age in tanks increases indefinitely

If the calculated water age continuously increases with time, that is an indication that the water that was present at that location at the beginning of the simulation is still there at the end of the simulation. Meaning, the water has not left the system. This can happen if for example the tank does not drain, or if the water parcels are simply shifting/sloshing back and forth. This situation can happen in cases where there is a long pipe connected to a tank, with fast filling/draining cycles. You can read more about this type of situation and more, in the troubleshooting water age article that Yashodhan mentioned.

Try graphing the tank percent full or hydraulic grade along with the related pump flow to check if the controls are working properly and to understand how quickly the tank is cycling. Adjust your pump controls as needed to avoid water age problems. Also, check your input data such as the tank size, operating range, node demands and review other model results to understand how the system is working, to determine why the water is not leaving the system.

Beyond that, please see the link in Yashodhan's reply for how to provide a copy of the model for further troubleshooting.

Jesse,

Thank you for your response. As I asked Yashodhan above, I noticed that when I set the tanks to 2-compartment mode. One of the tanks cycles around a consistent water age rather than increasing over time. When calculating the water age for a 2-compartment tank, does WaterCAD ignore the water age of the "dead zone" compartment?

Thanks,
Erica

Hello Erica,

This link has information on the different mixing methods available for water age. Some of this may be related to the set up of your system and your tank. I would set up the mixing method to best reflect how your tank is set up. If there are not two compartments in the tank, using this method may not be the best option, and using one of the other methods would be better.

If that doesn't help, seeing a copy of the model files may help us determine if the program is acting as expected.

Erica,

I see that you uploaded the model. I have taken a look and the issue you reported with the increasing water age appears to be working as designed. That is, I believe this is identifying a real problem with age in your distribution system. 

For my analysis I focused on "Glenview Tank". The travel time between the tank and the demands is greater than the cycle time of the pump, so there is old water "sloshing" back and forth. In my previous reply I had suspected this was the case, and this situation is mentioned in the following article in our Wiki: Troubleshooting Water Age Results

To explain - as the tank drains, water slowly makes it way out to the demands (between the tank and the pump) and by the time the tank drops to the level that causes the pump to turn back on, most of the water that came out of the tank has not left the system (via demands). 

When the pump turns back on, it quickly pushes all of that old water in the middle of the network back into the tank. It takes a relatively short time to fill the tank, but before it is full (at the point to turn the pump off), some new water does make its way to the tank. Since you have the mixing method set to LIFO (visualize as a tall skinny tank that fills/drains from the bottom), that old water goes up to the top of the tank, with the newer water from the pump at the bottom. Then as the tank starts draining again, it first drains the newer water at the bottom of the tank, so things look OK for some time, until the tank drains down to the point where the old water is, and then you see the "spike" in water age as it reflects the old water that has been "sloshing" between the tank and the network. That old water then slowly starts flowing out to the network (low velocities, high travel times, due to only 27 gpm total demand in that network), and the cycle repeats.

A good way to visualize this is to set up color coding on water age and animate or move the time slider around between the pump cycles (for example with a graph of tank level, inflow and age open at the same time), to see how the old water travels around. In Element Symbology, I set junctions to use GIS style symbology, with a size multiplier of 2. I used the following color coding on calculated age at junctions:

You will also want to set the initial water age of both tanks back to zero (one was set at 24 hr and the other 240 hours)

Lastly, you can take a look at the calculated Travel Time in the pipes when the tank is draining to see how slowly the old water moves back and forth.

To resolve this, you will need to adjust the operation of the tank and pump. There are several factors involved with this, such as the size of the pump (determining how fast the tank fills), the size of the tank (determining how quickly it drains), the demands (determining how fast the water leaves the system) and the pipe sizes (determining the velocity and travel time in the pipes).

With a total demand of only about 27 gpm, the 8" pipe sizes may be excessive, and lead to low velocity and thus high travel time.

As a side note, I noticed some seemingly conflicting controls for your pumps and tanks, which is leading to stability problems (notice how some timesteps in the calculation summary show a large number of trials, and at times I saw a few timesteps unbalanced). For example click the ellipsis button next to the Controls entry in the properties for pump FWP1 and review the controls.

Erica,

I see that you uploaded the model. I have taken a look and the issue you reported with the increasing water age appears to be working as designed. That is, I believe this is identifying a real problem with age in your distribution system. 

For my analysis I focused on "Glenview Tank". The travel time between the tank and the demands is greater than the cycle time of the pump, so there is old water "sloshing" back and forth. In my previous reply I had suspected this was the case, and this situation is mentioned in the following article in our Wiki: Troubleshooting Water Age Results

To explain - as the tank drains, water slowly makes it way out to the demands (between the tank and the pump) and by the time the tank drops to the level that causes the pump to turn back on, most of the water that came out of the tank has not left the system (via demands). 

When the pump turns back on, it quickly pushes all of that old water in the middle of the network back into the tank. It takes a relatively short time to fill the tank, but before it is full (at the point to turn the pump off), some new water does make its way to the tank. Since you have the mixing method set to LIFO (visualize as a tall skinny tank that fills/drains from the bottom), that old water goes up to the top of the tank, with the newer water from the pump at the bottom. Then as the tank starts draining again, it first drains the newer water at the bottom of the tank, so things look OK for some time, until the tank drains down to the point where the old water is, and then you see the "spike" in water age as it reflects the old water that has been "sloshing" between the tank and the network. That old water then slowly starts flowing out to the network (low velocities, high travel times, due to only 27 gpm total demand in that network), and the cycle repeats.

A good way to visualize this is to set up color coding on water age and animate or move the time slider around between the pump cycles (for example with a graph of tank level, inflow and age open at the same time), to see how the old water travels around. In Element Symbology, I set junctions to use GIS style symbology, with a size multiplier of 2. I used the following color coding on calculated age at junctions:

You will also want to set the initial water age of both tanks back to zero (one was set at 24 hr and the other 240 hours)

Lastly, you can take a look at the calculated Travel Time in the pipes when the tank is draining to see how slowly the old water moves back and forth.

To resolve this, you will need to adjust the operation of the tank and pump. There are several factors involved with this, such as the size of the pump (determining how fast the tank fills), the size of the tank (determining how quickly it drains), the demands (determining how fast the water leaves the system) and the pipe sizes (determining the velocity and travel time in the pipes).

With a total demand of only about 27 gpm, the 8" pipe sizes may be excessive, and lead to low velocity and thus high travel time.

As a side note, I noticed some seemingly conflicting controls for your pumps and tanks, which is leading to stability problems (notice how some timesteps in the calculation summary show a large number of trials, and at times I saw a few timesteps unbalanced). For example click the ellipsis button next to the Controls entry in the properties for pump FWP1 and review the controls.

Thank you this was extremely helpful! I really appreciate it!

Hi Jesse,

I noticed in the model that when the tank configuration is LIFO there are a couple locations where the water age decreases as it gets further from the tank. We would expect water age to increase as it gets further from the tanks. These locations are circled in red in the attached image. Is there an explanation for this? Would you mind taking a look at the model and letting me know what you think? Please let me know if I need to upload the model again. Thanks! PDF

Erica,

Most likely this is due to slugs of "new" water making it out to those locations. As seen in my previous reply, tracking water age can be quite complex. This is especially the case in a model where things are not in "steady state" as pumps turn on and off, flow reverses direction, and new water mixes with old water in complex ways.

To answer your particular question, you may need to closely examine an animation of the plan view with age color coding (not just the max age), to see where portions of newer water travel over time. You might find that new water travels back and forth as pumps turn on and off, but at some points a portion of the newer water is able to travel far enough so that it passes down a branch and toward a dead end, resulting in the situation you've observed. A smaller timestep may be required along with a finer granularity in your color coding, to track and observe the complex interactions.