We continue this series with the steps for designing control strategies. Simple rules of thumb will be offered for setting up cascade, composition, flow, level, pH, pressure, and temperature control systems. Examples of common unit operations will help provide understanding of the steps involved. The objective is to set the stage for tuning and configuration of the PID to do the best job possible.
- Add a flow measurement to every important process and utility stream to enable a secondary flow control loop for cascade control. A secondary flow loop isolates pressure disturbances, and nonlinearities of the installed characteristic of control valve and variable speed drives from the control of a higher level process variable such as composition and temperature. The flow measurement enables flow feedforward control and the possibility of changing continuous unit operation production rates by moving plant flows in unison per the Process Flow Diagram. The flow measurements also enable closing material and energy balances leading to better process knowledge eliminating uncertainties from pressure flow relationships. For info on the many benefits of flow measurements see InTech articles "Advances in flow and level measurement enhance process knowledge, control" (Mar/Apr 2011) and "Feedforward control enables flexible, sustainable manufacturing" (Jan/Feb 2010). A word of caution. Control valves and VSD normally have a greater rangeability than a differential head or vortex meter. When this occurs, a calculated flow based on the installed characteristic should be substituted for the measurement flow before the signal becomes too noisy or in the case of the vortex meter drops out. A bias to the pressure drop used in the flow calculation should be automatically set to provide a smooth transition from measured to computed flow.
- Realize that the input determining composition, pH and temperature loop response is a flow ratio. Plot these process variables versus the ratio of the manipulated flow to the feed flow. For vessel and particularly reactor temperature control, set up a secondary coil or jacket temperature loop to isolate utility system nonlinearities and disturbances from the primary temperature loop. Design the utility system to keep the coil and jacket flow constant to prevent a low heat transfer coefficient, fouling, and high dead time and high process gain at low flow. Use direct blending of utility streams and coil or jacket streams rather than heat exchanges to make the secondary loop faster. Avoid changes in phase in the coil and jacket. Use steam injectors to create hot water without bubbles eliminating the need to make a transition between cooling water and steam.
- Recognize that the input determining level and pressure loop response is the difference in flow entering and exiting the volume. Pressure control in general needs to be tight since pressure changes translate to flow changes in control valves and VSD and determine equilibrium relationships with temperature and affect the driving force for mass transfer (e.g. vaporization of volatile components and absorption of dissolved gases such as oxygen and carbon dioxide). Pressure also affects the reaction rate of gas reactants. Level control may or may not need to be tight. Tight level control may be necessary for tight residence time (volume/flow) in continuous reactors and crystallizers where the level setpoint changes with production rate to keep the ratio of volume to total feed flow constant. Tight level control is also important for recycle of distillate to column as reflux from a distillate receiver for self-regulation of reflux in the column and recycle of recovered reactants to a reactor self-regulation of concentrations in recycle system. On the other hand, many plant oscillations can be traced to a level control loop where tight level control was unnecessary and the level loop is excessively changing feeds to downstream equipment. The classic examples are surge tank levels and distillate level controllers manipulating distillate flow to downstream users rather than reflux to a column. As discussed in the March Control Talk Blog "Processes with no Steady State in PID Time Frame Tips (Conclusion)" simply decreasing the level controller gain without increasing the reset time may create slow oscillations.
- Choose the controlled variable that is affected the most by the process variable really trying to be controlled, typically a composition. For distillation columns, this rule corresponds to a tray temperature that shows the largest change to both an increase and decrease in the pertinent flow ratio. For acid or base concentration control, this rule translates to the use of pH for concentrations less than 0.1 normal and conductivity for higher concentrations.
- Choose the manipulated variable that causes the greatest effect on the controlled variable. For column temperature control a change in the reflux/feed ratio most often has a larger effect than a change in the steam/feed ratio. For columns separating extremely low concentrations of light (lower boiling point) components causing very low distillate flow, the distillate receiver level PID manipulates reflux flow to ensure the level loop can handle all operating conditions. For columns with separating exceptionally low concentrations of heavy (higher boiling point components), the sump level controller manipulates reboiler steam flow rather than bottom flow. Unfortunately this may introduce inverse response in the level control but if the bottom flow is too low to provide enough muscle to deal with range of operating conditions, the choice is manipulate to stream is right in that it always assures the ability to control level.
- Choose the manipulated variable that is least affected by disturbances given sufficient effect on the controlled variable. The classic example is the pressure and flow control of a pipeline with two control valves. The more important control variable is flow since this is what ultimately affects the downstream equipment. The flow loop should manipulate the control valve with the largest pressure drop so that pressure upsets have the least effect. This corresponds to the smallest valve (counter intuitive) given it can handle the maximum flow. Stroking either valve will provide the flow range so here the choice is determined by which valve minimizes the effect of common disturbances, which for flow is pressure. The pressure controller manipulates the larger valve. The pressure controller should be tuned for fast aggressive action as noted in step 8 while the flow controller should be tuned for slow smooth action. Pressure regulators did this naturally by providing high gain fast action. I prefer putting pressure measurement and control in the control room for visibility, adjustability, and diagnostics but the PID must have a fast execution rate and a high controller gain.
- Ask yourself the questions in the checklists given in the Sept-Oct 2012 Control Talk Blogs to make sure you covering all the bases for various types of control loops. For more much more extensive information checkout Advanced pH Measurement and Control and Advanced Temperature Measurement and Control
Next week we move on to configuring and tuning the PID controller.