Smart instruments are designed to help the maintenance staff at process plants anticipate equipment failures and prevent unscheduled process shutdowns. But that’s not the reality at most plants, where reactive, fix-it-when-it-breaks maintenance is still the norm. There are many factors contributing to this, from an installed base dominated by analog instruments to fears that change could create unacceptable risks.
While many companies turned to preventive routines based on historical experience to improve their maintenance practices, this strategy has also proven to be wasteful and costly. “Rather than relying on a reactive maintenance strategy, many process industries took the next step, which was to focus on time-based preventive strategies for devices,” explains Nicholas Meyer, product manager for reliability solutions at Emerson Process Management. “But that approach is labor-intensive and often results in excessive activity.”
Amit Ajmeri, asset management consultant for Yokogawa Corporation of America, cites studies that say 63 percent of field trips for routine maintenance work are actually unnecessary. “This unneeded maintenance spending,” he says, “represents a staggering 12 percent of manufacturing costs.”
Moving to predictive maintenance
Smart instruments are now coming into their own as maintenance practices shift from preventive to predictive strategies based on condition monitoring.
Not surprisingly, among the earliest adopters of predictive maintenance and smart instruments have been the industries where process shutdowns or catastrophic failures create the greatest risks to workers, plants and company reputations. These include the power and petrochemical industries, particularly refineries.
Moazzam Shamsi, director of global solution architects at Emerson Process Management, refers to a life sciences company where the facility manager implemented a radical change by eliminating time-based maintenance on most field devices.
“The company estimates a reduction of about 40 percent in the time required for instrument maintenance, saving approximately $52,500 per year. Applying predictive maintenance to the control valves was also economically beneficial,” he says. “By utilizing the valve diagnostics and pulling valves only when maintenance was actually needed, the facility began saving another $16,500 per year. The amount of time needed for the annual shutdown was also reduced by half and plant availability was increased, allowing the production of additional batches.”
Many barriers exist
With success stories like this, why do so many companies fail to fully utilize the capabilities of smart instruments?
“There are still many human barriers to the behavior changes needed to move to predictive maintenance,” Meyer says. “Company managers often mistakenly think this approach is too expensive. In reality, the 2010 NPRA Reliability and Maintenance Conference reported that reactive maintenance costs could be up to 50 percent greater than planned maintenance.”
Many plants also rely on a reactive mode because of workforce reductions and loss of expert knowledge. “Perhaps the biggest challenge,” Meyer adds, “lies in the culture shift and training needed to change work processes. Plants are concerned that they don’t have the time or the expertise to train people in a new way of working.”
Clinton Hommel, product marketing specialist, energy monitoring and management at Phoenix Contact, also points to communication technology as a major sticking point. “Though the use of familiar communication protocols has become more commonplace, a large percentage of devices can still only communicate on a proprietary protocol.”
The fact that most process plants are still living in an analog world presents another challenge. “When it comes to sensors, the 4-20 mA current loop is still the most common sensor output signal,” says Henry Menke, marketing manager for position sensors at Balluff. “Analog current loops have several advantages, such as simplicity, modest cost, easy troubleshooting and broken wire detection. Their disadvantages, however, include troublesome grounding and shielding issues, particularly on long runs, and distance limitations. These devices also typically deliver only their measured process data, limiting their usefulness in doing predictive maintenance.”
To be successful, predictive maintenance strategies require intelligent field devices, an open communication protocol, and integrated device and asset management software, Ajmeri says.
With instruments, particularly the valves that are critical to any process, getting smarter software that can manage all that data is becoming essential. “When a plant has 6,000 or even 16,000 instruments, what do you pay attention to first?” asks John Yingst, product manager for field device manager and fieldbus products at Honeywell. “It creates a kind of whack-a-mole environment. That’s why operators and facility managers need a better way to prioritize the information so they can make intelligent decisions.”
This potential for information overload is a major barrier to making better use of smart instruments, Hommel adds. “With all of the available information one can track on a process or machine, it’s really hard to tell from time to time what is important, and that information is sometimes lost in a sea of useless or non-critical information. When this occurs, most or all of the collection is ignored outright.”
Though vendors and fieldbus organizations may not have done a good job in the past in helping their customers prioritize device alerts, says Bart Winters, product director for asset management solutions at Honeywell, “that’s beginning to change.” Many vendors like Honeywell now offer instrument asset management solutions that sit above the device and DCS to provide an environment for decision-making.
Industry groups are getting involved and are proposing specifications for prioritization. One example is NE-107, which has been proposed by Namur, the European equivalent to ISA. In the past two or three years, prioritization has also been the subject of efforts by the Fieldbus Foundation (FF-912) and HART.
Vendors are actively working to develop prioritization tools. “You need to know the criticality of an instrument in the process and the severity of a fault so you can focus attention on what matters most,” Winters explains. “The goal is to make things like safety, security and alarm management more automatic. Instrument asset management solutions also keep a historical record so you can do root cause analyses.”
Vendors are developing process sensors with digital interfaces to improve the quantity and quality of sensor data, to make communication with sensors bi-directional and enhance ease of installation and troubleshooting, according to Menke. One such interface is IO-Link, supported by an industry consortium that includes Balluff. IO-Link operates over three wires (power, common and signal) at 24 V on unshielded cables, and is supported by any of the common industrial Ethernet networks.
“A smart digital sensor can not only deliver its process data, it can also report information like serial and model number for maintenance tracking,” Menke explains. “Some sensors will store information on an under-pressure or over-pressure event, for example, such as how low or high.”
He adds, “Many smart sensors can be remotely parameterized, so scaling of the measurement range can be done within the controller. This comes in very handy when replacing a damaged sensor. Rather than having to check calibration records and then bench calibrate or calibrate after installation, the configuration file stored in the controller is uploaded to the new sensor as soon as it is plugged into the control system.”
Wireless technology is also a tool for predictive maintenance. Emerson’s Meyer tells about a North American refinery that reduced the costs of pump repairs by using wireless temperature and pressure transmitters to alert operators when filters needed to be replaced. The wireless devices were installed at a 90 percent savings over traditional wired devices. By reducing unplanned pump failures, proactive monitoring helped the refinery increase the availability of its coking operation.
Expanding the infrastructure
Other industry groups, often based in Europe, are supporting standards and technologies to make it easier to apply smart instruments for predictive maintenance. The FDT Group AISBL, for example, promotes the open, vendor-independent FDT (Field Device Tool) standard to simplify the configuration of and access to field devices.
An FDT-compliant software tool called Device Type Manager (DTM), which manufacturers can apply to their devices, enables sensor data to be communicated across fieldbuses.
“Smart instruments that are supported by DTMs automatically provide proactive device health alerts,” explains Glenn Schulz, the FDT Group’s managing director. “Many DTMs also provide potential causes of a problem and the most likely corrective action, both for devices and their networks. When a network starts to exhibit noise, for example, the DTM will alert the operator to the condition, suggesting possible causes such as a lost termination resistor.”
The Profibus and Profinet International (PI) organization also supports manufacturing process control and asset management, providing an umbrella structure that enables online management tools to directly query smart devices for their current status. It covers diagnostic and alarm reporting by the instruments as part of their routine operation.
“The PI approach to predictive/proactive maintenance relies on the diagnostic/alarm schemes built into the instruments themselves. At calibration or commissioning time, the instruments are set with their operating parameters,” says John Swindall, director of the Profibus Test Lab at the Profi Interface Center. “This operational band is programmed in, such as allowable voltage levels for an analog input. Anything within that band is just data; anything outside the range can generate a diagnostic or an alarm that is displayed in the control room or on an HMI. As long as the people writing the HMIs and the control room applications know the data is out there, it can be accessed 24/7 by the people doing the maintenance.”
For some practical tips on predictive maintenance, see http://awgo.to/534.