What’s needed for a calibration analysis

A calibration analysis should consider how will the process operate and what is required to keep it running.

By Doug Taylor, Sam Cafferata and Ryan Beesley June 4, 2022
Custom vision systems can significantly benefit the customer by including an automated calibration feature. Courtesy: Concept Systems

 

Learning Objectives

  • Determine how to leverage manufacturer expertise when selecting or implementing a sensor to increase success.
  • Examine how disaster recovery should be considered in the event of a breakdown to reduce downtime and other potential issues.
  • Learn to consulting the sensor manufacturer and follow guidelines as a good starting point for implementation.

The precision in controlling a process is only as reliable as the data being reported; and the most important items in sensing are accuracy and repeatability. This is the case for manual and automatic control. Having properly calibrated sensors are key to understanding what is occurring. Sensor calibration can be a very involved process, but it should never be mysterious.  Some processes are very interdependent, and the process may be suffering from a variety of maladies: Improper tuning, incorrect accuracy dependencies, wear-related issues, or simple thermal drift – and it is easy to blame calibration for all the machine’s woes. Calibration issues are hard to diagnose because  often a mixture of issues contributes to the problem. 

What’s needed for sensor implementation? 

A key piece to an any sensor implementation into a process is the consideration of the calibration needs and requirements during system design. In the design and selection process, a calibration analysis should take place where there are two components to this analysis that need to be considered: The first is: how will the process operate, and what is required to keep it running? The second is consideration of any regulatory or commercial aspects to the calibration. Sensors used in any form of payment structure often have governmental or other calibration requirements. Weighing systems often have very strict certification and calibration requirements. For example, weigh scales measuring raw product with payment schedules are extremely sensitive and require regular calibration/verification. Fluid measurement often requires the same with gas delivery. Electrical metering is another area of intense scrutiny where certified utility grade devices are required.  

The first step in determining when a sensor needs calibration is understanding what calibration means and recognizing whether a sensor needs to be calibrated or parameterized. In many cases, the word calibration is used when parameterization of a sensor is a more accurate description. Process control sensors (level, pressure, temperature, flow, etc.) often come from the manufacturer calibrated for the process in which the sensor will be used. The manufacturer can provide a certificate of calibration, tracible to national and international standards and provides a reliable starting point when implementing a sensor.  

With a factory calibration, a sensor will only need to be parameterized to the physical attributes of the process on initial install.  

A typical example is a differential pressure sensor installed in a product tank to monitor liquid level. The tank characteristics will be applied to the transmitter, such as tank height, shape, product density, etc. and will further incorporate the sensor into the process. In some cases, it is necessary to “zero” the instrument due to elevation differences, environmental conditions or other factors, but this is not generally considered a calibration. 

Custom vision systems can significantly benefit the customer by including an automated calibration feature. Courtesy: Concept Systems

Custom vision systems can significantly benefit the customer by including an automated calibration feature. Courtesy: Concept Systems

In machine vision, robotic, and other applications, calibration might mean light sources are filtered and adjusted, or optical sensors tuned to measurement characteristics. Experience proves that calibration, parameterization, and development are an iterative process during integration. Typically, a sensor is rough calibrated, and then the process is broadly tuned. Then the sensor is calibrated to a higher level, which allows the process to be tuned to a higher state. This repetitive process continues until the machine is operating within specification.  

In process control, many factors can influence when a sensor needs recalibration: Installation environment, process material, measurement principle, etc. A sensor constantly exposed to caustic materials or severe temperature swings will require more maintenance and attention than a sensor in a more stable process.  

Guided-wave radar (GWR) sensors used in measuring wastewater effluent are more exposed to contamination and build up on the sensing membrane; whereas a submersible hydrostatic sensor is less susceptible and requires less attention. 

Fixed-mounted optical sensors often tend to not require recalibration until the unit fails and is replaced. Shock, temperature swings, and variable forces are not working against calibration, and reliability is often the result. Furthermore, designing-in features to protect for dust and debris effectively often yields a long and successful service life for the unit.  

Sensors mounted on moving constructs, such as robots, tend to have a shorter service life due to the mechanical trauma they routinely encounter. Cameras, lenses, and cables are very sensitive, so are often swapped out and recalibrated, but at a much higher total cost of ownership than the fixed-mounted units. 

Dealing with sensors that require regular attention 

Sensors operating at the far end of their accuracy envelope tend to be very sensitive to calibration and are at great risk of being problematic in a system, demanding regular attention and careful recalibration. Often, replacing these sensors is the first retrofit of a system once a more accurate sensor is available, usually incurring costs and downtime that far eclipse the original cost of the sensor. 

Other sensors that suffer from thermal drift include single-point triangulation lasers. When these sensors are operating close to their accuracy envelope, they usually require an automatic recalibration process where the machine simply recalibrates the sensor to overcome thermal drift on the fly. This requires the machine design to allows for sensor instability, and the process to afford the recalibration time. Systems that self-recalibrate can operate seamlessly and virtually without maintenance interaction when properly designed, even when replacing a component. It is often very expensive for a machine to recalibrate itself, but it can pay dividends in downtime recovery, a lower part-rejection stream, and a lower overall maintenance cost versus a manually recalibrated system that requires constant attention. 

Sensors operating on the edge 

In general, every sensor that operates close to the edge of its accuracy envelope experiences dynamic loading/shock, or the sensor has a lot of trauma from its setting. In these cases, regular calibration is required and is necessary to reduce downstream issues. On the other hand, fixed sensors well back from the process tend to not require recalibration. Experienced designers tend to favor locating sensors in safe locations when practical. 

Another aspect of any sensor is how vital is it to a process, and what is the disaster recovery plan. These should be developed for all critical and specialized sensors. What is the availability of a new one? If it must be changed out, the user needs to know whether a simple sensor change or a complete disassembly and reassembly of the tool is required. What does downtime cost, and what are the risks if the sensor goes down?  

This disaster recovery plan will help answer the question should the sensor be run to fail or replaced at timed intervals. This study can be very surprising and has led to the installation of an array of sensors for redundancy just to prevent down time and work-in-progress risk.  

Integrating sensors for monitoring 

Many things can be done on the integration side of the project to help keep track of sensors and how sensors are behaving. For example, a raw values trend can be set up, and the sensor’s performance can be compared to a data trace, or baseline, from when first implemented. Bounds can be set up to alarm sensor drift. Peak alarms also can be set to alert to possible sensor damage, such as a load being dropped on a scale. Using these techniques can improve overall user experience and the process reliability. 

Many emergent technologies enable maintenance teams, electrical and instrumentation technicians, and operations better insight into sensor health. Many sensors have internal integrity checks that can be enabled such that the sensor can alert on its own as well. Highway Addressable Remote Transducers (HART from FieldComm Group), EtherNet/IP (from ODVA) and other fieldbus communication-capable devices often put sensor health and diagnostics at users’ fingertips via software or specialized communication tools.  

Using these tools, a reliable preventive maintenance schedule can be constructed to help plan for scheduled calibrations and mitigate unplanned outages. Some manufacturers even market their own software to automatically monitor devices and alert the user of drifts in accuracy and build calibration schedules for the user. 

Protecting your sensor from the process, calibration 

Regardless of the sensor, having a well-documented calibration process that uses a durable and protected standard, it is often easy to use and store, which is the integrator’s responsibility. The integrator must provide a documented process and means to recalibrate the sensors being used. It is the responsibility of the customer to demand this upon delivery of the system.  

The most successful users record the process and provide laminated work instructions on or near the sensor guiding the technicians through the process. Unfortunately, many systems encountered are lacking in this important design element, and their processes always remain at risk for an unexpected and unwelcome process disruption. 

When implementing sensor technology, always consult the manufacturer on the application, maintenance, calibration requirements, and any special considerations when using their sensor. Not all processes are the same and leveraging the expertise of the manufacturer when selecting or implementing a sensor increases its rate of success. Applying these principles when using sensing technologies ensures the highest level of accuracy and repeatability for monitoring processes.   

Doug Taylor, principal controls engineer; Sam Cafferata, principal engineer; and Ryan Beesley, regional engineer manager, are with Concept Systems, a CSIA member; CSIA is a content partner. Edited by Chris Vavra, web content manager, Control Engineering, CFE Media and Technology, cvavra@cfemedia.com 

MORE ANSWERS 

Keywords: analytics, process sensors  

CONSIDER THIS 

Do you invest more in sensors and designs to reduce calibration requirements? 


Doug Taylor, Sam Cafferata and Ryan Beesley
Author Bio: Doug Taylor, principal controls engineer; Sam Cafferata, principal engineer; and Ryan Beesley, regional engineer manager, are with Concept Systems,