The biggest mistakes I have seen in the process industry came from trying to cut costs by not using the best technology instrumentation and systems. We tend to forget that measurements are the only windows into the process and controllers and final control elements (e.g., control valves) are the only means of affecting the process. If you can make the recognition of the essential role of automation part of your company’s culture, you can write your own ticket (e.g., trip to a plant).
In the days of pneumatic positioners, plants were flying blind so far as knowing what a control valve was actually doing. We didn’t realize that positioners out of calibration, poor actuator and positioner sensitivity, improper actuator bench settings, excessive friction, and linkage backlash meant that we were lucky to be within 5 percent of the desired valve position. A theoretical study in the 1970s said that boosters instead of positioners should be used on fast loops, not realizing the previous non-idealities or the fact that the booster set up a positive feedback loop that would cause a butterfly disc to become unstable. To this day, some companies still try to decide whether or not to spend a thousand dollars on a positioner, jeopardizing the loop and the process.
To more easily justify the cost of a DCS in the 1980s, thermocouple input cards were specified instead of temperature transmitters. The resolution of the wide range input cards of the 1980s and early 1990s was 0.25 degrees, preventing the use of the high controller gains and rate action on temperature loops. In addition, the individual drift of a thermocouple could not be compensated for and the individual error in percent of span could not be minimized by narrowing the calibration span per application requirements. Distillation columns and reactors that depended on tight temperature control suffered immensely.
The biggest mistake of yesterday (and still today) is using on-off valves as throttling control valves. Piston actuators, sloppy linkages, and spool type positioners are put on piping valves, making an attractively cheap valve with tight shutoff that meets the piping spec (see Tip #83). Packaged equipment vendors are always trying to be the lowest bidder. Many times any old thing that sort of works is OK. Common shortcuts include field switches and gages instead of transmitters, on-off valves instead of true control valves, orifice flowmeters instead of inline meters (vortex, magnetic, or Coriolis) and bare thermocouples instead of resistance temperature detectors (RTDs) in thermowells.
When onstream time and batches are worth millions, measurement life expectancies are less than a year, and/or drift can be greater than one percent per month, install triplicate measurement devices and use middle signal selection (see Tip #88). We triplicate devices in safety instrumentation systems, but fail to make the same cost-benefit analysis in process control systems.
Concept: Automation professionals need to be proactive and use their knowledge of the performance of the instrumentation and its impact on the process to provide systems that will exceed long-term process requirements. When in doubt, engineers should err on the high side. Future requirements in terms of flexibility and process knowledge for sustainable manufacturing are often underestimated.
Details: Specify instrumentation (measurements and final control elements) with the greatest reliability, lowest life cycle cost, highest turndown, least drift, least uncertainty, and best precision. For valves, the sensitivity component of precision is most important. For measurements, the repeatability component of precision is also important. The life cycle cost includes the cost of hardware, installation, piping, maintenance, and process variability. Choose a Distributed Control System (DCS) that offers the greatest I/O flexibility, most advanced control tools, best configuration capability, and PID features (see Tips #71, #72, #91, and #100).
For a ballpark estimate of the cost of process variability, take the square root of the sum of the squares of noise, repeatability, sensitivity, and uncertainty and ask a process engineer what is the impact on the process operation and process analysis. If you don’t have time to do the analysis, you can use the following guide: Estimate the life cycle costs of orifice meters as 20× the hardware cost, of vortex meters as 6× the hardware cost, of magmeters as 3× the hardware cost, and Coriolis meters as 2× the hardware cost. For reactors and columns, use RTDs for temperatures < 450°C. Don’t use on-off valves or field switches. Use 3 pH electrodes and middle signal selection (Tip #88).
Watch-Outs: The cost of impulse line winterization and maintenance, meter coefficient uncertainty, measurement noise, and reduced turndown is often not included in the life cycle cost of orifice meters. The cost of straight pipe runs and flow conditioners for orifice meters and vortex meters is often not considered because the cost is in the piping budget. The cost of lower sensitivity, greater drift, and greater risk for electromagnetic interference (EMI) of thermocouples compared to RTDs and the cost of extension wire are often not considered. The life cycle cost of field switches instead of transmitters and software switches often does not take into account the inability to verify their integrity and accuracy, their lower reliability and accuracy, and the loss of a control room signal for process analysis and troubleshooting. The life cycle cost estimates of control valves, dampers, and vanes do not include the process variability and maintenance cost of limit cycles and poor turndown from poor actuator and positioner sensitivity and total valve assembly backlash and stiction. Engineers who have not worked in a plant, who move on from project to project in an instrument design function (no onsite for checkout and start-up) and who do not have a dialog with plant operators, maintenance, and process engineers will mostly just understand hardware costs. These engineers have not seen the impact of instrument selection on operability, maintainability, and profitability.
Exceptions: Coriolis meters can get prohibitively expensive in large pipe sizes. RTDs have a much higher failure rate than thermocouples when vibration occurs, whether due to equipment, flashing, or velocity-induced vibration. Insulation resistance degradation at temperatures above 400°C can greatly increase RTD errors.
Insight: A focus on saving thousands of dollars up front can result in additional maintenance costs and can have hidden long-term costs of millions of dollars in terms of the loss in process efficiency, flexibility, and capacity.
Rule of Thumb: Do not go with cheaper instrumentation unless you are certain the impact on maintenance and process performance is (and will continue to be) negligible.