Process Control Improvement Recommendations Tips

Here we look at how to make sure the measurement system is able to provide the analysis, metrics, and control needed for process control improvement. Also outlined is the opportunity sizing and assessment process, the use of statistical tools, the tracking down of the source of variability, and the finding of more optimum setpoints.

Overview

An improvement in measurement accuracy enables more accurate material and energy balances, statistical analysis, and metrics. Routine verification of the calibration of sensors on-line and off-line can ensure a more reliable and effective measurement. The resolution, threshold sensitivity, accuracy and precision should be determined by measurement system analysis. The use of the latest transmitter and sensor technology can significantly reduce the need for recalibration. Some sensors offer much greater threshold sensitivity and sustained accuracy (e.g. less drift and installation effects) such as resistance temperature detectors instead of thermocouples and Coriolis meters instead of vortex meters.

The optimum setpoint can be found from historical data, response surface methods, simulation test results, and data analytics, confirmed by process analysis through design of experiments as to valid cause and effect. Unnecessary setpoint biases by operations are revealed in the differences in the performance of various shifts. The reasons for the perceived need for the comfort zone should be identified and addressed by better automation. Finding the optimums via historical data is more difficult for processes that run consistently because there is not enough change in the data.

Process and automation system performance can get better or worse after maintenance or changes in measurements, final control elements, process, piping design, configuration, and tuning. Periods of maintenance and changes should be noted and performance analyzed before and after.

Determine if you need to improve efficiency, flexibility or capacity. Efficiency and capacity could be both affected by flexibility if you consider that the rework produced during a change-over will affect your numbers and that during transitions you could be forced to decrease your current rate. Although you might be tempted to work in all of them at once, consider establishing priorities or get overwhelmed otherwise. If different teams are formed to work in parallel they might find each other affecting or interfering with others’ objective.

Opportunity sizing consists of finding the best periods of operation in terms of efficiency and capacity from simulations, historical data, online metrics, and cost sheets. These best periods represent the entitlement of the process or the desired target of operation. When compared with the current baseline the resulting gap represents an initial and potential window for improvement. The revenue minus the cost of goods can be computed for these periods for better analysis of efficiency, flexibility, and capacity tradeoffs. The gaps between the best and average plant performance is the sizing.

Once the objective and the opportunity sizing have been established, make sure the team is all onboard. In a meeting with key people from instrument and electrical, analyzer technology, quality assurance lab, process technology, research and development, operations, maintenance, configuration, and process control, identify possible solutions for each gap in the opportunity sizing. Build process control diagrams with relative locations shown in piping and equipment of all of the control loops, measurements, and final control elements (valves and variable speed drives). These diagrams should be done for utility systems as well. Provide online access to trend charts, troubleshooting results, online metrics, data analytics, statistical measures, power spectrum analyzers, and auto tuner or adaptive tuner results. For each possible solution, have the key process technology person estimate the portion of the gap that can be eliminated. A Cause and Effect Matrix could be used as an initial guide to each process variable that could affect the efficiency, flexibility or capacity. An exploratory graphical analysis using trend plots, main effect plots, CUSUM charts, and matrix plots can help evaluate the relationships.

Have the automation system people estimate the rough time and order of magnitude installed costs (total of hardware, design, and installation costs). Assign priorities and people based on the type of improvement and the possible benefits, time, and cost.

Oscillations are the principal causes of variability. The first thing to do is to put the subject loop in manual and see if the oscillations or are significantly reduced. If they stop the problem is in the loop tuning, measurement, of final control element. If the oscillation amplitude decreases, the subject loop is amplifying them.

When the culprit loop is found, an order of magnitude or more increase in the reset time can reveal whether the solution is tuning, less deadband, better resolution, or better threshold sensitivity. If the oscillations do not originate in a loop, then process instabilities, mechanical deficiencies, changes in phase and poor mixing are sources of fast oscillations. Recycle streams, batch operations, on-off control, safety instrumentation systems, relief devices, cyclic unit operations (defrosting and regeneration), tank car deliveries, and ambient conditions are likely causes of slow oscillations.

Performance must be analyzed preferably by online metrics immediately before and after process control improvements so that benefits can be properly assigned. There are many people making plant improvements anxious to take credit for benefits.

Recommendations

  1. Work with the automation system supplier (e.g. representative or local business partner) to estimate and improve installed accuracy of measurements.
  2. Use internal resources such as the Quality Assurance Lab to evaluate critical on-line and at-line measurements particularly those for composition and pH.
  3. Use historical data, simulation, design of experiments, surface response methods, data analytics, and process analysis to find optimum setpoints.
  4. Install online process metrics.
  5. Conduct an opportunity sizing and assessment.
  6. Use Cause and Effect Matrix to identify all inputs to the process that could be affecting efficiency, flexibility or capacity.
  7. Initially filter out those variables with minimum impact using exploratory graphical analysis.
  8. Use principal component analysis (PCA) and projection to latent structures (PLS) as part of a data analytics tool to find important correlations.
  9. Track down and eliminate or mitigate the sources of oscillations.
  10. Find and address reasons for offsets from the optimum introduced by operations.
  11. Implement basic and advanced regulatory control improvements.
  12. Improve setpoints and document and report benefits.

What are your comments?

Join the discussion today. Login Here.

Comments

No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments