Pumps & Systems, March 2007
Vibration monitoring of rotating equipment and analysis of the resulting data are effective ways to evaluate the health of production machinery in order to achieve best maintenance practices, extend equipment longevity, and avoid unplanned shutdowns. Plant equipment seldom fails without giving signals well in advance, so breakdowns can normally be predicted by listening for the warnings and passing that knowledge on to those in a position to prevent such problems.
Technology exists today to automatically obtain field-based information about the health and well-being of a whole range of critically important process industry machines. For example, turbines, generators, compressors, fans, motors, pumps and other rotating equipment can be continuously monitored online for changing vibration patterns and rising temperatures - sure signs of impending trouble.
However, most continuous monitoring systems alarm only when vibration becomes excessive, even violent. This kind of "trip protection" shuts down the machine without regard for the impact of that action on production. It protects the asset, but does not protect production goals.
Automated, full-time monitoring for in-depth information about the actual condition of a machine is generally reserved for equipment that, if stopped, would likely result in damage and shut down of all or a major section of the plant. This "most critical" category generally involves only about five percent of rotating assets. Data from automated monitoring systems enable plant personnel to predict when such a machine will need maintenance to prevent damage and lost production.
Just below the "most critical" category is another group of machines that would adversely impact productivity by more than 40 percent if they fail. Pumps moving vital fluids through essential processes fit into this tier of importance, which might be called the "next frontier" for automated monitoring. Today, vibration levels of next frontier pumping systems are generally checked periodically using route-based data collectors, but the collection is rarely continuous and provides only a "snapshot."
This is beginning to change, with new systems emerging for automated data collection and analysis in the field. When combined with information from other sources, such as lubricant analysis and infrared imaging, a true picture emerges of the operating condition of monitored assets and their potential for failure, providing the basis for an effective predictive maintenance program.
Integration with Process Control
Automated process control systems and machine monitoring systems are very different, having been designed for use by two different groups within a plant. These standalone systems have existed in the same buildings for years, but have rarely been integrated. However, the importance of bringing together data relating to the health of machinery in the top tiers is apparent.
According to Ron Moore in his book, Making Common Sense Common Practice, 40 to 50 percent of equipment breakdowns are related to poor operating practices. Mechanical failure is listed as the number one cause of large losses in process industry plants. It seems obvious that if the operators knew more about the condition of the equipment under their control, they would better understand how their actions contribute to mechanical issues.
For example, if an improperly operating or specified relief valve on a compressor causes back pressure fluctuations, a mechanical failure is nearly inevitable. By recognizing a concurrent increase in machine vibration, operators should be able to get at the root of the problem and initiate corrective action to keep the process operating.
Integration of machine health data on the most critical machines and "the new frontier" of critical machines with the main control system will lead to the following plant wide benefits:
- Reduced impact of process on machine health
- Fewer false alarms
- Better and earlier diagnosis of machine problems
- Better planning and scheduling of routine shutdowns based on predictive intelligence, leading to faster, more effective maintenance/repairs
Even though monitoring vibration and controlling process parameters require different kinds of hardware with dissimilar scan rates and uncommon methods of applying collected data, the combination of these two systems can unleash power and benefits without the need for new technology. The two systems can be integrated without a great deal of additional expertise.
For example, when a health monitor overlays an existing protection system for advanced online condition monitoring, the continuously generated data can be integrated in at least three ways:
- Results are provided to the Maintenance Department where a health manager suite evaluates the data, assigns a severity ranking to existing conditions, and issues early warnings, along with diagnostic suggestions as to the root cause.
- Live vibration information is fed directly to operators, giving them a real-time view of the effects of their actions and enabling them to exercise control schemes that will best preserve the production machines. Warnings and diagnostic information can automatically show up in the plant's control system or come in the form of an e-mail.
- Information summarizing and prioritizing ongoing alerts and events is transmitted via a portal for viewing by plant management for an overview of asset health at any level.
Where It's Working
Integration of information from unmanned remote pumping stations to the SCADA host at Tarrant Regional Water District in Texas has resulted in early identification of such problems as unbalance, misalignment, looseness, cavitation, and gear and bearing faults. "Run-to-failure is no longer an acceptable option - not when continuous, on-line condition monitoring is available to trigger reliability-centered maintenance," one Tarrant official said. "It is just too costly to let those large, unmanned pumps run until something breaks."
In addition, keeping tanks at optimum levels is now based on a new process parameter called vibration. According to the Tarrant official, "We found that as tank levels drop, vibration increases, and this decreases the life of the pump over the long term. Since we have control of the tank levels, it is easy to feed this machine health characteristic back to process control where the levels can be adjusted accordingly."
As a result of the availability of this kind of information, unexpected critical mechanical failures have been minimized, emergency costs have been reduced 20 percent, and staffing requirements for new pumping stations in the growing district were cut in half.
Conclusions
The goal of integration is to get the right information to the right personnel in order to get more out of production systems without unexpected emergencies and downtime. Operations personnel are often asked to push the limits of the processes they control (speeds, loads, pressures, etc.), but they won't get there if they can't see the impact of those actions on the mechanical systems.
Bottom line impact will be seen by integrating field-generated machinery health data with control systems and reflected by continuous improvement of the production process.