By Dan Hebert, PE
To some, condition monitoring (CM) means installing vibration sensors on a few critical motors, and then having a maintenance technician with a handheld data logger check sensor readings on a more or less regular basis.
To others, such as Romel Bhullar, senior technical fellow/director at the engineering and construction giant Fluor (www.fluor.com), CM is a very big deal, requiring a systems approach. "At Fluor, working as systems integrators on mega projects, we routinely incorporate conditioning monitoring solutions to control and manage capital assets," says Bhullar. (Figure 1)
Figure 1: When Fluor designs and builds huge process plants for its clients, the condition-monitoring system is a key deliverable.
Photo courtesy of Fluor
Condition monitoring to keep modern process plants up and running are necessarily complex. "A conditioning monitoring project typically involves integrating and marrying diverse and competing commercial technologies, systems and parameters," explains Bhullar. "It also requires very high-speed, real-time input monitoring, signal conditioning and processing. We have to deal with different sensor suppliers, vibration systems and system integration problems—using a variety of networks, communication protocols, media, servers, and application hardware and software."
However, it pays off, especially in process industry applications, where plant shutdowns due to equipment failure can cost millions of dollars per day. The dollars and the challenges may seem daunting, but the rewards can be significant, and there are ways to cut the project down to a manageable size.
Getting Started
While a plant-wide CM and asset management system isn't a simple project, the approach to it can be fairly straightforward, according to Michael Gurney, co-CEO of Concept Systems (www.conceptsystemsinc.com), a systems integrator in Albany, Ore.
"When it comes to the actual implementation of a CM solution, there are two options," he advises. "The first is using control hardware already in place, and the second is equipping assets with new instrumentation and controls."
Much hardware installed in a plant already has diagnostic capabilities. "In virtually all cases we've looked at, there are capabilities already on the plant floor for capturing data on an asset, but most go underutilized," says Gurney.
Sometimes, you don't even have to install sensors, as everything you need may already be there. InduSoft (www.indusoft.com), for example, installed a SCADA system at the University of Texas in Austin. "The system monitors more than 1200 meters from more than 100 buildings on the campus," explains Fabio Terezinho, InduSoft's VP of Consulting Services. "The InduSoft system combines real-time data acquired from an OPC server with historical data retrieved from the historian, and displays information from electrical, chilled water, domestic water and steam on dashboards—and makes the information available to remote users anywhere via a web browser."
The system also generates reports with a log of events, such as calibration/verification of the meters, which is used to schedule preventive maintenance and check the integrity of the data. Alarm and remote notification allows the university to detect and fix defects in a timely manner. According to InduSoft, installation went smoothly. "Integration of the open architecture InduSoft Web Studio with third-party systems previously installed at the site was straightforward," says Terezinho.
Another example of a relatively easy condition monitoring implementation occurred at Centro Energia Teverola (www.centroenergia.it), a 150-MW combined-cycle cogeneration power plant in Teverola, Italy, about 20 miles north of Naples. The plant already had CM sensors installed on instruments and valves, all wired into the existing legacy control system. But it was experiencing unexpected performance degradation because of clogged filters on the gas turbines, and the legacy system wasn't helping.
To obtain the advanced analysis needed to identify the problems, Centro Energia installed Emerson Process Management's (www.emersonprocess.com) AMS Performance Monitor. Emerson wrote a macro to acquire the necessary data from the legacy control system's historian. (Figure 2)
Figure 2: Centro Energia's 150-MW, combined-cycle cogeneration power plant in Teverola, Italy, is monitored remotely by Emerson Process Management in the U.K.
Photo courtesy of Emerson Process Management.
Improving Existing CM Systems
Some plants may already have a conditioning monitoring system that isn't doing the job. Vectren (www.vectren.com), a power company in southwest Indiana, wanted to monitor more than 500 existing assets. Vectren's engineers were using a combination of software and spreadsheets to analyze more than 15000 equipment readings. The trend analysis software checked for problems and reported instances when assets were broken or not running efficiently. Alarms indicated problems that needed immediate attention, but the existing approach had several drawbacks.
"Dangerous conditions could evolve quickly and undetected, increasing the likelihood of costly unplanned outages," notes Isauro Martinez-Cairo, director at Invensys Operations Management (www.invensys.com) "The monitoring procedures were time-consuming, and the cause of problems couldn't always be determined immediately. Alarms were also repetitive and didn't reflect problems that a combination of conditions might have triggered."
Invensys helped Vectren achieve a better solution by implementing Avantis Condition Manager, an automated CM software solution. It tracks the performance of assets to determine exactly when maintenance should be performed, generating alerts as conditions are triggered. Staff receive warnings when conditions begin to degrade, enabling them to perform maintenance as needed, rather than as dictated by an arbitrary schedule. This reduces unplanned downtime, and enables a highly effective predictive and proactive maintenance strategy.
"The preventive maintenance functionality alone provides enough savings to pay for the entire implementation," reports Scott Brown, a reliability engineer at Vectren. "Instead of wasting an engineer's time reviewing charts, that engineer is freed to do other tasks."
Marathon Oil (www.marathon.com) had a similar problem. Marathon operates the East Brae gas platform in the North Sea, 165 miles northeast of Aberdeen, Scotland, and it had an aging vibration monitoring system that monitored and reported on the condition of the equipment. The former system was rapidly becoming obsolete, so Marathon turned to Rockwell Automation (www.rockwellautomation.com) for an upgrade.
Figure 3: Marathon Oil gathers data with Rockwell Automation's intelligent I/O modules to monitor the operation of gas compressors on this platform in the North Sea.
Photo courtesy of Rockwell Automation
Advanced Monitoring
We've discussed a few examples of companies replacing older CM and manual systems, and the rush may be on. As asset management and CM hardware and software become less expensive and easier to use, it pays companies to invest in the technology.
For example, in our April 2010 cover story (www.controlglobal.com/articles/2010/WorkersUnchained1004.html), we discussed how wireless technology makes it easy to add assets to a CM system and eliminates the need for operators to walk around the plant with a data logger.
Most of the major process control system vendors understand the need for CM and offer packages that work with their control systems. Or, as we saw in the Centro Energia case, their software works with other vendors' new and legacy systems as well.
Some exotic technology is available to solve tough problems. Tony Amato, president of Swantech (www.swantech.cwfc.com), Falls Church, Va., a systems integrator and supplier of CM technology, explains how one company diagnosed problems with a stress wave energy (SWE) analyzer.
"The company detected increased stress wave energy on a bottoms pump for a distillation column. The system had been in place for more than a year, and the SWE had been trending low and consistent during that time. By correlating increases in SWE to historical process data, the customer found the increases in SWE coincided with decreased and/or erratic pump flow. Further analyzing the differential pressure and the column level indicated the problem was fouling in the column packing or trays, leading to increased stress on the bottoms pump. The column was inspected and fouling was present. Following cleaning, SWE returned to normal levels," explains Amato.
Gurney of Concept Systems says companies should consider two underused technologies: statistical process control (SPC) and digital video recorders (DVRs). "SPC tools offer better data analysis and earlier, more accurate detection of problems," he explains. "Small variances, oftentimes operating within normal high/low limits, reveal equipment issues well before catastrophic failure."
As for DVRs, Gurney says they offer the ultimate troubleshooting device by giving personnel the ability to record processing lines when upset conditions or downtime are detected. "This way, plant personnel can go back and look at the video captured during these events and determine the root cause of the problem," he says.
As CM capabilities increase and costs come down, the challenge for many will be finding the time to install and operate these systems. Despite proven payback, plant personnel often are so burdened with day-to-day operation that they can't find the time to use proven CM tools.
Lack of skilled personnel to analyze and interpret condition monitoring data is another continuing challenge. A solution may be off-site analysis of CM data by third-party suppliers, with actionable data provided to plant personnel on an exception basis as is the case with Centro Energia's use of Emerson's off-site analysis services.
State-of-the-art CM is still not easy or inexpensive, but it's also not an insurmountable challenge, and the expense and effort are well-spent when weighed against the cost of downtime for critical assets.
Dan Hebert is Control's senior technical editor.
Condition Monitoring Tells All
Fluor designs and builds huge process facilities such as chemical plants, refineries and power plants. Romel Bhullar, senior technical fellow/director at Fluor, says, "In almost every project we do—refineries, energy, petrochemical, biochemical, pharmaceutical or power—asset condition monitoring (CM) is an integral part."
One such project at Potomac Electric Power Co. (Pepco) involved a plant supplying power to the city of Washington, D.C. "It's critical that the power generation turbines, generators and auxiliary systems have a CM system that monitors impending failures and identifies reasons to do preventive maintenance."
Putting together such a system isn't easy. "Standard commercial off-the-shelf solutions weren't flexible and functional enough to match our needs," he says. "There were mountains of challenges: technical, people, commercial, multiple company cultures and organizations," he adds. "In these days of tough economic times, all vendors are protecting their markets and providing proprietary and expensive solutions. It makes integration very difficult."
Things are improving, but problems still exist when tying together different communication protocols. "There have been significant developments in networking technologies and communications equipment, but major issues in the bottom two layers of the OSI model still need lot of work," he notes. "Management may have different objectives than the people in operations, IT and control systems—not to mention suppliers. It's a maze out there." Bhullar says the cost to install the CM system at Potomac Power "…was several million dollars."
The data being gathered is extremely important to more than just the power company. EPRI (www.epri.org), a power industry consortium in Palo Alto, Calif., contributed to the project, so it could see the data.
"The conditioning monitoring system provides data in real time to all the EPRI stakeholders across the U.S.," Bhullar explains. "GE is looking at their new gas turbine/generator and collecting performance parameters to improve their design. Westinghouse/Mitsubishi is collecting data on their newest gas turbines/generators. Potomac Power's operations and maintenance staff are looking at impending failures. And Fluor is collecting data for validating our methods for predicting failures."
The system works, too. "The cooling system on one of the turbine blades had failed due to plugging of the passages," he says. "The system identified the problem, and Potomac was able to replace the blade in off peak hours without any loss in power."
It works for EPRI too. The information is being used by member utilities to make decisions relating to equipment selection, design configurations, maintenance and replacement policies. The data is also being used by these same utilities to validate supplier sales claims and promises of newer technologies before making the huge required capital outlays for a comprehensive CM system.
Attacking the Problem
Systems integrator Concept Systems in Albany, Ore., has implemented many high-level asset condition monitoring (CM) solutions. Michael Gurney, co-CEO, says that a CM project takes a two-level approach: First, use control hardware already in place; next, equip assets with new instrumentation and controls.
Much of the equipment in a plant already has diagnostic capabilities. "A typical automation system will control the equipment, but stop short of pulling data off the controller or motor drives. With some simple programming, a system can collect key data such as motor start/stops, cylinder cycles, fault counts, drive current, downtime, cycle time/flow rates, valve position, pressure readings and more," says Gurney.
"Collecting and correlating this data finds problems," he explains. "For example, monitoring the flow rate at known valve positions will point to issues with the valve prior to catastrophic failure. All this information can be used to make better sense of the data and better manage that asset."
Some equipment may not have the sensors needed for CM. In this case, the asset has to be equipped with sensors and tied into the control system. "The possibilities are really endless and depend on the type of asset being monitored," points out Gurney. "Because of this, a crucial step is evaluating what assets really need to be monitored and how. The question is what assets put my processing line at greatest risk. In many cases this may not be known, and that is where data gathered from existing control hardware can help."
Critical areas where a CM solution could be usefully employed include: mechanical systems for vibration and current monitoring; pneumatic systems for temperature, pressure, condensation, air flow and filter monitoring; hydraulic systems for oil pressure, temperature, level, flow, accumulator and filter monitoring; and electrical systems for power quality and temperature monitoring.
Gurney says several options exist for analyzing data. "In many cases, data can be fed directly into a computerized maintenance management software system, and the data never needs to be directly presented to maintenance personnel," he says. "Companies will also offer to take on this function for customers—where they analyze the data directly and work with the maintenance staff or with third-party mechanical, pneumatic, hydraulic or electrical specialists."
The other critical element of a CM solution is measuring the results. "Overall equipment effectiveness determines gains recognized by a CM solution," he explains. "Having this data prior to the CM implementation will deliver quantifiable data on the improvements when viewed before and after the project."
Powerful Condition Monitoring
The installation at Centro Energia Teverola's 150-MW combined cycle cogeneration power plant in Teverola, Italy, near Naples, illustrates two key aspects of modern condition monitoring (CM) and asset management systems.
First, it may not matter what kind of new or legacy control system is installed at your plant because modern CM software packages can work with almost any system. Second, you don't have to develop specific knowledge about CM because outside services exist that will analyze the data and recommend best courses of action.
Centro Energia had a legacy Bailey Infi-90 control system with 5000 I/O, and it wanted to monitor clogging problems with inlet filters on its gas turbines as well as other operations. The data Centro Energia needed to analyze was already being collected by the control system, so all the company had to do was install Emerson Process Management's AMS Performance Monitor software and contract for Emerson's remote analysis services.
It's necessary to monitor the gradual deterioration in filter performance and calculate the cost of the resulting reduction in turbine performance. By comparing this with the cost of the maintenance required, the most appropriate point to replace the blocked filter can be determined.
But Centro Energia doesn't make the determination: Filter and other data from Teverola is transmitted to Emerson's performance monitoring center of expertise in Teesside, in northeastern England, where experienced engineers analyze the data and produce reports showing performance. These reports can be accessed via any standard web browser, and are based on thermodynamic models developed for each machine.
In addition to the online information, Emerson also provides advice about the operational efficiency of machinery. Existing or potential problems are highlighted, as well as new opportunities to improve overall efficiency.
Vincenzo Piscitelli, general manager, Centro Energia Teverola, says, "We can assess the effectiveness and economic return of our maintenance activities, which allows us to determine what maintenance work is required to improve equipment performance."
This improved planning has also enabled Centro Energia Teverola to reduce average repair times from seven hours to two hours.
