The Art of Instrumentation

At a recent process industry user group meeting, I was asking engineers informally if they believe their processes are adequately instrumented. Answers varied, but most suggested they could probably use more process data. One dissenter related a report that he’d heard of: A well-known oil refiner had categorized the criticality of instrumentation devices within the plant.

By Peter Welander, Control Engineering November 1, 2008

At a recent process industry user group meeting, I was asking engineers informally if they believe their processes are adequately instrumented. Answers varied, but most suggested they could probably use more process data. One dissenter related a report that he’d heard of: A well-known oil refiner had categorized the criticality of instrumentation devices within the plant. Some were classified as highly critical, while the other end of the spectrum included a group designated, “if it breaks, don’t replace it.”

Instrumentation devices often collect other types of data in addition to a primary process variable. Often this is used internally to provide some type of correction factor, but HART communication or a fieldbus can provide the means to extract the other data when it is useful. This diagram (right) shows how a Coriolis flowmeter (left) uses data collected by direct measurement to calculate additional values. Photo Source: Micro Motion

There is such a thing as too much information, but that is probably, at worst, either a waste of resources or an annoyance to operators. On the other hand, having too little data on process performance and equipment conditions can reduce effectiveness and even be unsafe. New technologies enable wider deployment of devices than ever before, but the decisions are still more art than science—and highly application specific.

“It all comes down to process control design,” says Praveen Muniyappa, chemical and biofuels business developer for Siemens Energy & Automation. “There are certain times where there’s a lot of instrumentation, but they don’t use all that information. But more often processes are under-instrumented, and there’s a requirement for customers to go in and add parameters, whether its level for safety, flow for process control, temperature, and so on. They want to add instrumentation, but they have to live with existing constraints.”

When engineers are laying out a greenfield process unit, or considering a major equipment upgrade, there is a critical balance between broad instrumentation deployment with its high cost, and taking a Spartan approach that minimizes device counts. “When EPCs (engineering, procurement, and construction) are designing processes, certain measurements are called out so that you can control the vessels and processes to get the right results out of the end,” says Brian Dickson, vice president of field device business for Invensys Process Systems. “What happens is, in the vetting process, much of the instrumentation is sliced-and-diced out of a design if it’s proven that it doesn’t offer enough value to justify itself. The cost per point has certainly gone down over the last 15 or 20 years, but it still costs a lot of money to build a plant, so a lot of the instrumentation desired from a design point of view gets cut out of the package. We do find, very often, that another measurement here or there would offer a lot of value.”

Wireless, fieldbus, and other networking platforms can reduce installation costs, or at least lower wiring costs. Still, it is important to consider all the costs and benefits involved when making a business case for adding new instrumentation.

Quantifying costs, benefits

There are countless circumstances that drive a company to add more devices to a process or supporting equipment. Common scenarios might involve:

Upgrading a process control platform to add advanced process control or an optimization utility that requires additional data;

New safety requirements which call for additional level or pressure monitoring points; and

Adding conditioning monitoring to rotating equipment to assist maintenance.

Any such projects should have objectives with definable financial goals or compliance drivers to weigh against the full range of associated costs. Situations where operators want to add a single or small group of devices may be harder to quantify, at least on the benefit side.

When considering adding devices, it’s important to include all the real costs rather than understating the full range of expenses. For example:

Engineering for planing and documentation;

Device purchase price;

Physical installation and wiring;

Integrate new data into the DCS; and

Ongoing maintenance.

The extent of these individual cost components depends on the sophistication of the installation. Removing a

Identifying benefits

One automation consultant for a large oil producer, who asked not to be named, is not so sure adding condition monitoring sensors to equipment always pays for itself. Since most process-critical installations are generally in a tandem back-up situation, is it really worse to run a pump until it stops? He observed that when the pump goes out, the oil producer simply switches to the other one and repairs the motor. In the consultant’s opinion, if an installation is running properly, adding monitors doesn’t really save anything, because they neither extend the life of the motor nor reduce the repair cost.

But such situations are not always the case. There will be times when you have no choice but to add new devices. Upgrading process control strategy or adding new safety functions may call for more data points, and that can drive the implementation decision. For example:

Dickson says that when he does pinch analysis at a refinery, he has to add temperature measurement points to the process. “We can solve two problems if we use a conductivity sensor, because conductivity has temperature as part of its process,” he says. “It gives our measurement for the analysis, and we have a conductivity measurement to give us a diagnostic on the heat exchanger itself. If it has cracked tubes, conductivity changes and we can give an alarm. As a result, we’ve positively affected both the process and maintenance. One tube failure caught early can be worth hundreds of thousands of dollars.”

Tim Wortley global product manager for GE Sensing sees different drivers. “In life sciences industries, it’s really hard to change a process. If they need to pull in extra data, it’s usually because of a regulatory need to do that. Someone says, ‘You need to measure this process parameter here because we say you have to.’ It’s not because it’s going to bring some amount of savings. Everyone is moving toward process efficiency, but the biggest driver is still regulatory compliance,” he says.

More, or smarter devices?

Most instrumentation devices made over the last decade have HART-enabled transmitters. While this supports specific diagnostic functions, there are often secondary variables that can be helpful. For example, many types of pressure sensors require internal temperature compensation and therefore have that measurement capability. That variable can typically be accessed through the HART data.

Gary Cusick, fieldbus marketing manager for Invensys Process Systems observes, “It’s all about the capabilities to get more information from current instrumentation. What’s really driving growth of Foundation Fieldbus and Profibus PA and WirelessHART is the type and quality of information you can get from a physical device installed in the facility. Multi-variable sensors provide more data from one device, which is one reason you don’t need more devices, just another type of device, and because of digital technologies, you can have higher density devices. One of the biggest reasons we see people upgrading to digital bus technology is the intelligence in the devices. So when you look at the lifecycle of a plant where the maintenance people take down a piece of equipment and rebuild it on a scheduled periodic basis, now they can run it until the device tells them ‘I need maintenance,’ extending the period between maintenance, and reducing maintenance parts and labor.”

Using that information does not necessarily require more sophisticated field communication protocols. “Most I/O communication is 4-20 mA, but we see a trend where users are going to HART I/O with that capability built in,” says Siemens’ Muniyappa. “They can take traditional instruments and get all that HART data that they could not bring in before. But if a legacy system can only take 4-20, the only alternative is to take a multiplexer or interface that can take the HART signal and break it up into multiple parameters. Let’s say you have a tank with two liquids where one is in a layer on top of the other. With a TDR [time domain reflectometry guided wave radar] level sensor, you can measure the top level in the tank and the interface level with only one device. The main 4-20 signal measures the top level, and the HART signal can provide a secondary 4-20 for the interface level measurement.”

While these secondary variables can be considered free data, their usefulness depends on being able to get what you want and at the place you want it. Returning to our earlier example, the temperature from a pressure transmitter may be of the process fluid or it may be of the sensor, depending on the transducer configuration. Secondary variables are chosen first to support the primary variable. Manufacturers offer different options. For example, pressure sensors will not always offer temperature. The most important thing is understanding what that data is and how frequently it is updated. When that is clear, you can use it appropriately.

“Using secondary variables from HART devices for any kind of control in a DCS can be ‘iffy,’” says John Yingst, product manager for asset management, Honeywell Process Solutions. “Yes, it can be done, but be sure to understand where the variables come from and how meaningful they are. Does the temperature of a pressure transmitter represent the fluid or the ambient surroundings or something in between? With a valve actuator, do you know which variable represents the actual valve position? Not all actuators are alike, so you definitely want to test this. Also, most HART I/O for a DCS is multiplexed, so there is no guarantee of bandwidth, especially when someone has a calibrator connected to the device. It is important to test bandwidth thoroughly before committing to any kind of critical control. Despite these cautions, though, some customers who know what they’re doing say they have successfully used secondary variables in control logic.”

Footprints in the process

Implementing new instrumentation requires physical mounting, communication infrastructure, and supporting engineering efforts. Fieldbus protocols and wireless instrumentation have reduced the cost of communication while increasing the volume and sophistication of data carried. Control system and instrumentation vendors for the most part are working to make this as simple and painless as possible.

“A lot of control system vendors are making HART data capture very easy on the graphic interface or HMI side,” says Muniyappa. “They now have pre-built templates where the user doesn’t have to know what the different HART parameters are, since they have a pre-built screen that shows the tag name, the model number, when it was inserted, what’s the minimum range, what’s the maximum range, all in a pre-built screen so somebody doesn’t have to work through all the logic to address all those HART parameters.”

Still, physically installing a device and its ancillary equipment can get costly. Muniyappa adds, “If you want to get a level or pressure measurement way out on a tank farm, the cable cost could be a couple thousand dollars, and the labor cost could be that amount. So the fact that the transmitter is only $2,000, the total cost, including engineering, design, documentation, could be five to six times that amount to get the variable back into the system. We can look at a specific application and recommend a radar level sensor that has a two-inch diameter horn. But if the only available nozzle on the tank is one-and-a-half-inch, we now have to find a technology that will work with the existing conditions and still get the best performance. For the customer to shut a tank down to change or add a new nozzle could be $10,000 to $20,000. We, as an instrument vendor, think it’s easy to add an instrument, but for the end customer it’s a big deal.”

In some cases where a plant is being run by an older control system, the ability to add data points may not be so easy. “Often legacy systems cannot accept additional points in their systems,” says Scott Saunders, vice president for Moore Industries. “If the logical point count is nearing its ceiling, adding more I/O will further take a toll on the existing scan rates used for control. In these situations moving the control overhead to the field with new smart I/O devices can get you increased I/O capacity without degrading your core control system’s performance. These concentrator devices can have a real-time control engine to do control, alarming, math and other functions in the field, for a fraction of the cost.”

Ultimately the need to add new devices or gather more data from what is already there depends on the characteristics of a specific process and what you have to do to make it operate reliably and economically. A process unit that is stable and operates consistently can get by with the least data. But when things begin to go wrong, more information can help troubleshoot upsets. Those are the situations where operators wish they had more information. Like most aspects of running a plant, the better you know and understand your process, the better off you’ll be.

Author Information

Peter Welander is process industries editor. Reach him at .