Process Control Improvement Perspective Tips
The benefits from process control improvement originate from increases in process efficiency, flexibility, and capacity. Often, there is a tradeoff where an increase in flexibility or capacity is accompanied by a decrease in efficiency. Better measurements, process knowledge, online metrics, and an enhanced PID can provide a more intelligent and effective optimization that minimizes the tradeoff between benefits.
Process knowledge improves as measurement accuracy is increased. Simulations are better verified, material and energy balances closed, better setpoints found, and more accurate metrics computed. The door is then opened to more effective process control improvement. The variability can be decreased as much as necessary by better basic and advanced regulator control. Operator biases (comfort zones) can be eliminated by automated protection to enable the setting of setpoints closer to the optimum. The setpoint proximity to the optimum can be optimized by override and valve position control. The benefits can be ball parked first as an opportunity sizing and then estimated by an opportunity assessment. The documenting and reporting of the actual benefits gained enables support for future opportunities. To see how Monsanto used these techniques to achieve an average 4% reduction in the cost of goods, see the June 2012 Control Talk Column “The Human Factor”.
You cannot control what you don’t measure. Online metrics are essential for developing, maintaining and justifying process control improvement. Nothing speaks as powerfully as money. Online process metrics are essential for focus, understanding, implementation, achievement, and recognition by the automation engineer and the profession.
Efficiency is the largest factor in the variable costs for the cost of goods sold (COGS) in manufacturing processes. The efficiency in the use of each raw material and utility stream for each unit operation should be computed online and made available to the operator real time. Normally this is done on a basis of the mass of each raw material and the energy of each utility stream used per unit mass of product. For batch processes, mass and energy use are totaled for each batch. Batch energy use is normally much less of a concern than capacity as indicated by batch cycle time. For continuous processes, the use is computed on a flow rate basis often requiring some intelligent filtering and synchronization of input and output flows to eliminate noise and a confusing irregular or inverse response.
Attaining and maintaining optimum operating conditions requires knowing the optimum setpoints and pushing the actual setpoints as close as possible to the optimum based on operability and variability. The optimum can change with production rate, raw materials and impurity, feed stock, product type and grade, recycle stream, ambient conditions, and the operating conditions for the process and utility systems. The efficiency during startup and transitions (changes in feedstock, product type, and product grade) should be computed and included in hourly, shift, daily, weekly, and monthly averages.
Flexibility corresponds to the ability to quickly and efficiently change production rate, feed stock, product type, and product grade based on market demand. The goal to minimize inventory makes flexibility more important and more difficult.
Capacity is affected by production rate and on-stream time. The production rate for batch processes depends upon batch cycle time and wait time. For continuous processes, the final product flow is the production rate. Start-up time, downtime, and transition time undermine on-stream time. The production rate for all types of operation should be computed and included in hourly, shift, daily, weekly, and monthly averages.
A valve position controller (VPC) can maximize production rate or minimize energy use by maximizing a feed rate or coolant supply temperature or minimizing a compressor discharge pressure. The main problem is the complexity of the tuning to eliminate interaction between the VPC and process loops and the resulting slow VPC response to upsets. The key here an enhanced PID for the VPC to suppress oscillations, make tuning simpler, and enable move suppression via setpoint rate limits on the PID being manipulated by the VPC for smooth optimization and fast disturbance rejection.
To enable greater flexibility in meeting different production rates, feedforward flow control should be used. Part of this realization is the recognition nearly all loops responsible for maintaining stream conditions (e.g. composition and temperature) are a function of the ratio of manipulated flow to feed flow and all loops responsible for inventory (e.g. liquid level and gas pressure) are a function of the difference between the manipulated flow (e.g. exit flow) and the counteracting flow (e.g. entrance flow) for the given volume. Keeping the flow ratio constant for stream conditions and keeping the entrance and exit flows equal for inventories are the keys to tight control. Flow feedforward is the most underrated advanced regulatory control technique. Also, the need to startup on flow ratio control and provide the operator the ability to see the actual ratio and correct the ratio setpoint has somehow been lost. The interfaces and functionality developed to provide this functionality in the 1980s need to be standard DCS features. For details on this missed opportunity see “Tip #93: Use a Viewable and Adjustable Flow Ratio and a Feedforward Summer” posted June 18, 2012 on the ISA Interchange site.
The frequency and amplitude of oscillations are signatures of different problems. Statistical methods can provide the distribution of amplitudes and power spectrum analyzers can provide the relative contribution at particular frequencies. These tools should be used to analyze violations of constraints and the results included in shift, daily, weekly, and monthly averages with the corresponding frequency noted.
The sources of oscillations can often be tracked down based on oscillation frequency (oscillation period) compared to the subject loop’s natural frequency (ultimate period). Often a surge tank level controller tuned too aggressively is the culprit. The other major sources of oscillations are loops with too much reset action (too small of a reset time) and valves with excessive backlash or stiction. With insight as to candidates for sources of the oscillations based on an oscillation frequency, straight forward troubleshooting techniques should be used such as putting individual loops in manual until the oscillation stops. If oscillations persist with loops in manual then the environment, equipment, or process are the usual suspects.
The attenuation of oscillations by intervening volumes should be estimated. The ratio of the output to input amplitude is inversely proportional to a fraction of the volume residence time (volume/flow). For well mixed volumes the fraction can be taken as one.