Lower-cost machinery health monitors are beginning to be offered. The difference between these and the more comprehensive ones, such as the CSI or GE Sensing versions, is generally that there’s less intelligence embedded in the lower-cost units. This is leading to two basic models for machinery monitoring. One, typified by the Emerson CSI units, puts intelligence at the device level, and only data that has already been reduced goes to the control system or the online asset management system. In the other, raw data is sent to the control system, which performs the analyses there.
Data Into Information?
So, we’ve gotten to the point where we all agree that predictive maintenance is a good thing, and that there are ways to automate data collection. How do we get from there to actually having the data in a location and format that makes it meaningful, and therefore, <ital>information.<end ital> That’s the problem.
The other problem is how to propose a project to your management without saying that you’ll have to put a lot of equipment, hardware, and software in at a large startup cost before you start saving a single dollar from predictive maintenance.
HART: The Missing Link
The HART protocol is installed in 20 million devices worldwide. That’s a huge installed base. Most people use the protocol for configuring and troubleshooting individual devices, and not much else. But that’s not the only benefit you can get from the HART devices you already have in your plant.
In 2004, the HART Plant of the Year Award went to the Cooper River facility then owned by BP. A.J. Lambert, instrumentation and electrical reliability specialist, reported then, “Before using HART to connect our field devices to our asset management system, we had a major shutdown every two years, and pulled out 30 to 50 valves. Now with more information, we pull maybe five or six valves, and we know exactly why we pull them.”
How does he know? His HART data, tied to his asset management system allows him to see the projected failure rates for his valves and controls.
“We’ve realized over $2 million in documented savings to date,” said Johan Claassen, E/I manager for Sasol Solvents, of Sasolberg, South Africa, the 2005 HART Plant of the Year winner, when accepting his award, “and will continue to enhance our use of HART and expand on its benefits in the future.”
There are literally dozens of companies offering some sort of software for asset management, from packages like Flowserve’s Flowstar.net to offerings from Invensys, Honeywell, Siemens, Rockwell Automation, and Emerson’s AMS product, which we’ve already discussed. Web-based Flowstar.net is interesting because it includes “an active partnering process” with Flowstar.net’s tool to help plant personnel gain the most use from the tool.
In other words, plants using Flowstar.net outsource their asset management functions to Flowserve’s personnel. As we’ve reported before (See “The Art of Asset Management,” CONTROL, Nov. ’04, and “Serving Up Asset Management,” CONTROL, Nov. ’05) there is a movement that’s captured the interest of overworked plant executives and hungry automation vendors alike to outsource maintenance and repair services to a vendor organization. Most large vendors, such as Rockwell, Emerson, Flowserve and others, offer this service to a greater or less degree. Such sophisticated asset management software, including Honeywell’s PKS Asset Manager, Siemens’ PDM, Yokogawa’s Plant Resource Manager and others, can even be used remotely through OPC.
There are even software packages, like Matrikon’s OPCVigilant, that are device manufacturer and fieldbus type agnostic. As long as the asset you’re trying to manage has an OPC driver built into its code, OPCVigilant, and any other OPC enabled software product, can extract the data from that asset, and assemble it in a form that can be used by nearly every control system and middleware (MES) vendor in the world.
Selling Asset Management to The Boss
A.J. Lambert suggests that you start small, like he did. His investment of approximately $10,000 generated more than $300,000 in documented savin gs by the end of the first year.
Experts like Lambert point to the ubiquity of HART devices, and suggest that you begin by identifying a device or set of devices with a high probability of failure. Start by grouping those assets into an asset management system. Once the payback from your initial project is proven, it should become easier to gain approval to add more devices to the asset management system. With the advent of HART Wireless (See Sidebar Story below), it will become even easier to do simple, small projects, and network them together one at a time.
Once you have the basic management of your assets under control, and you’re using predictive maintenance, it’s time to look beyond the device level, and see if you can better manage groups of assets. The next step is predictive maintenance based on the control loop as the asset to be managed, and beyond that, the process as a whole, considering itself as an asset.
Alireza Haji-Valizadeh, technology development manager at ControlSoft, says it’s extremely useful to treat the control loop itself as the asset, rather than the devices that make up the control loop. Once you start thinking that way, he suggests, it becomes rather easy to extend the concept to the entire plant.