In our prior post, we talked about using variants of the classic PID algorithm in order to control vacuum processes. In the simplest form, the PID control loop can be reduced to a P-only, or 2-point, algorithm. For example, by using a PLC or other control module to read an accurate vacuum sensor and actuate an appropriately sized solenoid valve, a proportional-only control scheme can be readily implemented.
In this scheme, the solenoid valve is often plumbed directly into the vacuum line, between the process (e.g., a distillation column or drying oven) and the vacuum pump. The vacuum sensor would be placed as close as possible to the process. As distillate is pumped off, the absolute pressure will fall. Once the pressure drops below the set point, the valve will close. The pressure will then begin to rise as more vapor is produced. And as the pressure level rises above the set point, the valve will once again open. This cycle is repeated until the process is complete, as represented on the schematic below.
Actual Pressure vs. Time with a P-only Control Algorithm
In some cases – if, for example, the vacuum pump has a relatively large pumping speed as compared to the system volume – it can be helpful to add a bleed valve to the control system. A pump with relatively large pumping speed can cause the actual pressure to go well under the set point, resulting in poor process control and lower yield. For this reason, adding a bleed valve – which admits air or an inert gas if the pressure level drops too low – can mitigate this overshooting problem.
Whether a bleed valve is incorporated into the control system or not, the algorithm must be tuned in order to achieve an appropriate balance between holding the process pressure close to the set value without cycling the control valve too quickly. The algorithm is tuned simply by adjusting the proportional gain value in the control loop. If the actual process pressure differs too greatly from the set point, the process will not be well-controlled. When the control loop responds slowly so that the actual pressure does vary widely from the set point, it is indicative of a proportional gain that is too low. On the other hand, trying to hold the process pressure too close to the set point can result in the solenoid valve being opened and closed rapidly. After a number of cycles, the valve can fail and require replacement or repair. This sort of fast or aggressive response from the control loop suggests a gain value that is too high. Extensive resources are freely available online which discuss PID loop tuning in further detail, including information about determining the optimal proportional gain value.
Though in some cases it may be desirable to develop a vacuum control system and implement a PID algorithm, there are numerous off-the-shelf vacuum controllers which enable precise control and require no in-house development effort. These controllers often have more advanced capabilities available than a “home made” controller. For example, off-the-shelf units can typically be programmed to control the process according to a desired pressure-time profile, in addition to being able to hold a static set point. Off-the-shelf models often also include preset tuning features that allow for the control algorithm to be tweaked in order to optimize process control without a need for the user to become an expert in control theory. Some controllers, such as VACUUBRAND’s CVC 3000 detect controller, also include advanced features designed specifically for vacuum control applications in specialty chemical and pharmaceutical industries. The CVC 3000 detect is capable of automatically sensing solvent evaporation and maintaining the vacuum at this particular level. This provides a level of automation to process development, while also preventing sample loss or contamination that can occur during drying processes like distillation or work in drying ovens when process control is inadequate.
If you’d like to discuss how VACUUBRAND’s control capabilities might help to improve your process, please contact us.