Cooling Tower Staging

[flukeynub]

Is it normal to have chilled water plants that do not stage up a 2nd cooling tower before a 2nd chiller?

I would expect that having two cooling towers (2 towers 1 pump = 1/2 flow to each tower) at the same flow rate would increase the latent heat rejection to atmosphere and prolong the stage up of a 2nd chiller.

 

[Shophound]

It depends on how the towers are controlled. If they are controlled by condenser supply water temperature, the protocol may be to stage on/ramp up or stage off/ramp down towers as required to maintain the condenser water supply temperature setpoint.

The setpoint itself could be reset by outdoor wet bulb, as is the case in certain plant optimization strategies. I like to see towers operated that keep centrifugal chillers comfortably away from their surge envelopes, personally.

 

[flukeynub]

You would be correct. Our two towers are controlled via floating setpoint with WB as the process variable. We are supposed to have chiller redundancy (N+1 1250 ton chillers) but cooling loads have increased to a point where we reguarly stage a second chiller on during summer days.

Why do you try to prevent the chiller from surging? I'm still in the learning process and wish to know more about chiller operations.

 

[Shophound]

I run my towers the same way, resetting condenser water supply temp via wet bulb approach, ~ seven degrees for our Evapco towers. That said, I have limits on the protocol in place so there is a maximum and minimum setpoint. I've also found I needed to work a buffer into the program to slow reaction time to changes in wet bulb, otherwise the frequency drives on the tower motors hunted excessively.

Chiller surge is bad. A centrifugal compressor is a non-positive displacement pump. If the pressure difference becomes too high between the condenser and the cooler (evaporator), gas flow reverses across the impeller, at the same time attempting to stall the impeller. This causes a huge jump in compressor motor amps and also strains the drivetrain of the motor to impeller power transfer. On the higher efficiency chillers, tolerances between the impeller and its surrounding enclosure is very small; unchecked surging can lead to increased bearing play and gear lash, which in turn can lead to the impeller striking a stationary surface. Not good!

Many chillers of more recent vintage have surge prevention algorithms. I've set mine a bit conservative so it will go into surge prevention mode quicker when operating conditions deteriorate to that point. Conditions favorable for surge to occur is high entering condenser water temperature, not to mention clogged tower and condenser water pump strainers, fouled tubes due to scaling and/or debris, or a fouled tower. Good water chemical treatment and annual punching of condenser tubes are vital to keep a chiller healthy and away from surge territory, along with good tower management.

 

[KiwiMace]

This also depends on how well your towers respond to having 50% flow in the hot water basin. Not all towers run well like this. Talk to your vendor.

 

[flukeynub]

Ah, great explaination. Do you use a current limit to prevent surging? Or does the PLC "throttle," for lack of a better term, chiller capacity to prevent this high pressure differential?

Either way, sounds like it's time for me to speak with my Trane rep/engineers for some more information on both parts of the system.

 

[Shophound]

My chillers are Carrier 19XR. The surge prevention algorithm looks at four points: difference between entering and leaving chilled water temp (delta T) along with pressure difference between cooler and condenser at low load (delta P), and the same parameters at max load. When these two plot points operate outside the normal range determined by the above parameters, the chiller's computer will hold the guide vanes to prevent the chiller from loading up more. It may also close the guide vanes incrementally if the computer does not see improvement in the monitored parameters within a given time frame.

This chiller also has a way to detect how many times the compressor surges, in case the prevention algorithm can't react strongly enough to prevent a surge. Every time a surge occurs there is also a concurrent surge of amp draw, which the computer monitors via current sensors on the main power feed to the compressor. When a preset count is exceeded, the chiller trips and the compressor stops.

Additionally, one can select a "demand limit", which limits how high the chiller can be loaded. This is measured in percentage of FLA.

I'm curious as to how what I've described above compares to how different brands of chillers, such as Trane in your case, control to prevent surging from happening.

 

[Shophound]

KiwiMace also has a good point; some towers with reduced water flow may experience "vortexing" in the sumps (given that pumping capacity is the same but has merely been distributed between more than one tower cell), which is that mini-tornado of water you'd see as a kid when you let the water drain from the bathtub. Of course that depends on tower configuration.

My own towers are upflow forced draft and don't have that problem, but do have an overflow issue if the tower staging and floats are not set correctly. This is because an equalizer line separate from the pump suction was not installed when the towers were new. We plan to change that sad fact at some point.