Getting stronger process readings

Getting an accurate process variable measurement depends on many factors, but it must begin with the right instrumentation.

By Megan Wiens November 29, 2018

Consider this typical situation: a process unit in a chemical plant needs a pressure reading from a reactor. The unit’s automation system uses the reading to help determine reaction rate, and the operators in the control room want to see it on their human-machine interface (HMI) screen. If the reading is lost, the process must be shut down. How should the plant’s instrumentation engineer make sure the reading is correct and always available to the operators and the automation system?

For the purposes of this article, the focus is on measuring pressure, but the concepts apply across much of the range of process instrumentation. Think for a moment about what must happen to get the reading from one end of the chain to the other:

  • The process fluid must have a path to reach the sensor.
  • The pressure sensor must be deflected in a measurable way in proportion to the fluid pressure against atmosphere.
  • An electrical element must convert the deflection into an analog signal.
  • The transmitter must clean up the raw signal, scale it, and transmit it using a prescribed format, either analog or digital.
  • Wires, or perhaps radio, must carry the signal to the input/output (I/O) card of the automation system.
  • The automation system must convert the signal into appropriate engineering units for display and use the data in whatever equations it is performing for the larger control effort.

All of those elements must work correctly to capture the right value in real time to support control and monitoring. Let’s pull on the chain a little and look for potential weak links.

Through to the sensor

The sensor itself is the one moving part of the entire picture. It must flex in response to the pressure in a predictable and consistent manner. Usually, it involves a metallic isolator diaphragm with a strain gauge, or the movement is measured through capacitance (see Figure 1). For a strain gauge, movement must be linear enough to deflect the same amount for the same degree of pressure change, and return to its original form when the pressure is relieved. It must be sensitive, but at the same time able to withstand pressures far higher than its reading range.

For the sensor to be displaced, there must be a mechanism to allow the process fluid to press against it, which means there must be a process connection. The process fluid must have a path to the sensor, or the pressure gets transmitted via an uncompressible intermediate fluid, in which case there will be two isolator diaphragms with a fill fluid trapped in between. There also may be an impulse line so the sensor can be mounted some distance from the actual process equipment.

The path to the sensor or isolator diaphragm must be clear and unimpeded. Any blockage can slow response or decrease accuracy. Some transmitters have the ability to determine when an impulse line is plugging and can alert operators of the problem forming. Slugs of gas in a liquid line, or vice-versa, can cause reading inaccuracy. Impulse lines should be clear and bled, although it can be beneficial to have some condensate in a steam pressure line.

The physical and mechanical elements of the reading chain are pretty well understood and established. Sensor technology has not changed much over recent years, but there have been more dramatic advances in the next stages.

The busy transmitter

The transmitter of a process instrument has a lot to do, so much that we tend to refer to a complete instrument as a transmitter. Technically though, the transmitter is a specific part of the larger unit, with the sensor being the other main part.

The transmitter takes the raw analog signal from the sensor and cleans it up. Depending on the type of sensor, it might generate a resistance, voltage, capacitance, or some other type of signal in response to the process variable. The transmitter must measure the signal and compare it against the desired range according to how the device is configured.

This can be an involved process (see Figure 2). For example, the transmitter often needs an internal temperature sensor because the output of the pressure sensor is affected by temperature in addition to pressure. Additional corrections must be made for any nonlinearity of the sensor. All of these elements, and more, must enter into the signal processing.

The transmitter might even have a local display to show the primary variable in appropriate engineering units, and perhaps secondary variables such as temperature. Ultimately, the final output is the result of calculations and/or look-up tables to scale the reading according to the transmitter’s configuration. The development of more sophisticated electronics has made this much easier, often without users recognizing the improvements.

This in itself is impressive, but only part of the picture. Today’s transmitters have additional capabilities such as self-diagnostics. While the main signal processing functions are going on, there also are internal functions evaluating the quality of power coming into the transmitter, the condition of the internal electronic components, sensor functionality, and other items, including plugged impulse lines as just mentioned. If a problem is present or developing, the diagnostics can send a warning via a digital wired or wireless protocol.

Getting to the I/O point

Somewhere the processed electronic signal must be turned into something a human being can understand. If the reactor pressure is 110 psi, it must be converted and displayed as that value or 7.59 bar or whatever the operators need. The local display on the transmitter is handy for assisting with configuration and troubleshooting, but there are few applications where a human will be close enough to the transmitter to see it on a regular basis and react as required (see Figure 3).

Some sort of automation system invariably enters into the picture. It may be something as basic as a remote display with simple alarming functions, or a more elaborate platform such as a programmable logic controller (PLC) or a distributed control system (DCS). Traditionally, this involves wiring to transmit a 4-20 mA analog plus HART signal or digital fieldbus data, but WirelessHART now is being used quite frequently.

For traditional users with the wired analog approach, there are different ways in which the automation system can process what it’s getting from the transmitter. Some can read HART natively, but if the I/O only has the ability to read the analog signal alone, all the other diagnostic information will be stranded in the transmitter. It can warn of problems, but the system won’t be able to hear it.

One way to address this issue is by adding a WirelessHART converter to the transmitter. This allows the 4-20 mA signal with HART imposed to still be transmitted to the automation system, with an additional WirelessHART signal transmitted elsewhere, such as an asset management system, with all diagnostic information intact.

Some of the most sophisticated diagnostic capabilities extend beyond the device itself and examine the larger control loop. Where smart I/O is available, the diagnostics can detect and report conditions capable of distorting the reading and creating a false value. This warns operators of situations that appear to be correct but are wrong.

As straightforward as wiring is, it often emerges as the weakest link. In a traditional wired environment, the path from the actual transmitter to the I/O card of the DCS easily can include multiple marshalling cabinets, hundreds of feet of cable and perhaps 20 terminations. If the wiring is 15 to 25 years old, which is certainly not uncommon in today’s facilities, the insulation is probably getting brittle and terminal strips are quietly oxidizing. If left undisturbed, these connections can work for a long time, but if cable bundles are moved around to solve a problem or if marshalling cabinets are opened for troubleshooting, problems can develop. For example, shorts and signal drop-outs can occur, disrupting the control room and automation system.

Fortunately, new wiring and I/O systems can drastically reduce the number of terminations while adding native HART connectivity. Moreover, using WirelessHART can reduce the terminations to perhaps two or even zero. Particularly when working in legacy environments, WirelessHART can have a higher reliability rating than wired networks.

Sum of the parts

As can be seen, all the elements must work together for effective measurements. Any part performing badly can impede communication or cause it to break down altogether. Some companies try to solve the situation by treating the symptoms rather than root cause. If a critical value can’t get through to the DCS, the solution may be adding a redundant instrument on the same faulty infrastructure. It might improve the chances of data getting through, but it is like a Band-Aid.

This is not always the best approach, but sometimes it is a practical one. However, adding a redundant path via WirelessHART is a much better option. If it isn’t possible to upgrade existing infrastructure, simply adding another path subject to the same problems is not the best option. Even if there is no existing WirelessHART network, a small cluster of wireless transmitters can establish a sustainable network and support a variety of instruments that do not depend on legacy wiring, or even need to interface with the DCS using conventional I/O cards (see Figure 4).

Strengthening the links

It’s important to know the weak links of a system, but building strong links is just as important. In many respects, having the best possible transmitter is the most important element. If the source of data is not reliable and accurate, the best infrastructure in the world won’t make it any better.

The instrumentation available today from a variety of suppliers is sophisticated, stable, accurate, and reliable. If used in an environment capable of interacting with it fully, it can provide secondary variables and diagnostic information in addition to its primary variable.

As companies rely more on automation in a growing variety of ways, reliable instrumentation is foundational.

Retro design with today’s sophistication

Many users like the functionality of a traditional pressure gauge, but not its fragile Bourdon tube, finicky mechanism, and instability. Some wireless pressure gauges use a sophisticated sensor with transmitter electronics to provide the capabilities of a full electronic transmitter, but in an analog gauge form factor with a traditional needle display (see Figure 5). A small stepper motor moves the needle in real time. Add basic diagnostic functions and WirelessHART communication, and it checks the most-wanted feature boxes without the problems of traditional pressure gauges.

Megan Wiens is a global pressure product engineer for Emerson Automation Solutions in Shakopee, Minn., where she is responsible for advanced diagnostic capabilities across the Rosemount pressure portfolio. In this role, she works to implement product solutions that improve plant safety, increase process efficiency, and enhance process insight.

This article appears in the Applied Automation supplement for Control Engineering and Plant Engineering.

– See other articles from the supplement below.


Author Bio: Global pressure product engineer, Emerson Automation Solutions