Monitoring Cooling Towers

Over the years there has been no shortage of recommendations for monitoring cooling towers. If you ask four people you can get as many as half a dozen suggestions. One thing for sure is you can put too much monitoring on a cooling tower if you let yourself be talked into it. Few users want to wait for failures before they fix problems with motors, couplings, drive shafts, gears, bearings or even blades. Nor do they want to perform maintenance on a time based schedule. This paper reviews practical approaches to having a dependable monitoring capability as well as provide signal sources for a periodic monitoring program.

Scheduling Maintenance
Scheduling maintenance, based on machinery condition, is being done on major plant machinery today by many users. Many plants now include balance of plant (BOP) machinery in such programs. Cooling towers are included, in varying degrees, in periodic vibration monitoring programs designed to alert the user of changes in running conditions which might be the start of potential problems. These can appear as increasing trends in the vibration level or as significant changes in the vibration "signature".

A vibration monitoring program that includes periodically measuring vibration on the drive motor but not the gear, because it is not accessible, is good but not as complete as it could be. Many users are installing vibration sensors, usually accelerometers, on cooling tower gear assemblies and running the interconnect cable out to an area that is accessible by personnel. The cable is usually terminated in a connection center (or switch box) where it is out of the weather as well as the sometimes harsh environment. These sensors remain unpowered until the data collection instrument is connected to them. Power comes from the data collector. Needless to say, a periodic inspection program can be enhanced by adding measurement points on the gear this way.

Continuous Monitoring
Continuous monitoring is a good idea for cooling towers because some types of failures occur suddenly. The most common type of sudden failure is broken blades (part or whole). Others are usually structurally related. Mechanical drive system problems can be seen, most of the time, in a periodic monitoring program, before they become critical.

Most cooling towers were originally equipped with mechanical switches (commonly refereed to as "kick" switches or "earthquake" switches). These are often tied into the motor drive power circuit and will shut down a cell in a tower if activated, offering protection against catastrophic damage. Often the vibration switch will not respond when it should for a variety of reasons. For example, if a switch is mounted on a wooden cross member that has loosened, it may not respond to a loose coupling bolt even if it causes an unbalance condition. The vibration switch does not offer any protection in this case. Recognizing this fault, many users have replaced these devices.

A continuous monitoring system, the kind used to monitor critical machinery, offers several advantages for monitoring cooling towers. One advantage is the sensors can be mounted directly on the bearing support parts on the motor and the gear so bearing condition, as well as casing vibration can be measured. Also, two levels of alarm can be set so operators can be made aware of potential problems at early stages of development before a condition becomes critical. The second level of alarm can be used for shutdown in the same manner as did the vibration switch. The problem with conventional continuous monitoring is simply that it costs too much and is hard to justify.

A Reasonable Cost Solution

A modular continuous monitoring system, such as the Hardy Instruments DI 5500 Series, offers low cost "building blocks" that are ideally suited to monitoring cooling towers. The modules include vibration detection modules, dual set point modules, and relay modules. Other modules are available but these three types can be combined to provide a very reliable continuous monitoring capability for cooling towers.

A minimum system, per cell, is illustrated in Figure 2. Two accelerometers are used, one on the motor inboard bearing and one on the gear near the input shaft bearing. Both are in the horizontal plane. Two signal conditioners are used , one for each sensor, which convert the sensor signal to a velocity vibration, and provide output drive signals to a single dual set point module on an "auctioneering" basis. That is, the highest signal is the signal the set point module responds to. The outputs of the set point module are used to drive one or two relay modules, depending on whether or not two alarm levels are needed. The modules are mounted on a din rail inside a stainless steel enclosure which is installed up near the machinery or down at ground level. A power supply module is also required which operates on 120 Vac power.

In this configuration continuous monitoring and protection is provided for both the drive motor and gear assembly at a minimum level. This is a good system! Here is where it can get expensive! It is easy to say "lets put on some additional points"! Additional points are a good idea from a vibration analysts stand point but they do not necessarily make a better protection system. But lets review them anyway. Alternate configurations would include a second accelerometer on the outboard bearing of the motor and a second, or even a third, accelerometer on the gear. The second accelerometer would be mounted on the input shaft bearing in the vertical direction and the third accelerometer would be mounted on the back side of the gear housing in the axial direction, as illustrated in Figure 3.

The user has the additional benefit of being able to include these points in a periodic monitoring program where trends are easily observed. These trends consist of overall vibration, vibration spectrum, HFD and Envelope measurements, for example. The accelerometer makes HFD and Envelope measurements possible which offers the user alternate methods of tracking the condition of bearings and gears. The signal conditioners have a buffered signal output that allows for easy connection of analysis instruments.

System Expansion

Another output signal is provided on all Hardy signal conditioners. A 4-20 mA output signal that is proportional to full scale is standard. This signal can be used as an input for indicators, chart recorders, a PLC or a DCS system to log and display vibration levels as well as to trend them. Most of these data logging instruments can be configured with alarming functions as well.

General Installation Concerns

A cooling tower is a unique piece of plant equipment and require some special consideration when installing and planning a monitoring system. First is grounds. It is important to make sure there is a good electrical ground on both the drive motor and the gear assembly. Letting the vibration sensor return lines be the only ground source will get you a noisey monitoring system for sure! You can imagine what 10 mV of noise will look like on a 4-20 mA loop! Typically we tie the sensor shields to earth ground but keep the signal conditioner common isolated from earth ground through the power supply.

The next item where cooling towers are different from most of the other machinery in the plant is the alarm philosophy. Do not use high speed rotating machinery alarm concepts for cooling towers. If you do you will get a lot on nuisance alarms! Cooling towers shake! Set the alarm levels high enough. Numbers like 0.5 ips for alert and 0.8 ips for trip would be unheard of for high speed machinery but they are common on cooling tower gear assemblies. Also, set the alarm delay time long enough. Do not be afraid to set 6 to 8 seconds. You will avoid a lot of false trips and you will not jeopardize the monitoring and protection concept. Think of it this way. If you lost a blade, you would not do significantly more damage in 8 seconds than you would in 3 seconds. The benefit is, when you do get an alarm, you can be confident there is something wrong.

This article is provided by Hardy Instruments, www.hardyinst.com.