Combining several schemes for heat exchangers control, like feedback, cascade and feedforward control, in an integrated approach, has proven to be an excellent strategy to address the control requirements of even highly challenging heat exchanger applications.
Heat Exchanger Control Strategies – Integrating Feedback, Feedforward and Cascade Control
This article is aiming to present an overview of the various approaches of heat exchanger control, without diving into mathematics or differential equations, yet providing solid understanding of the different schemes implemented in the industry. This understanding forms essential knowledge, yet of extreme significance, to process control engineers.
Shell and Tube Heat Exchanger at a Glance
By far, shell and tube is the most common type of heat exchangers used in petrochemical industries as it is suitable for low and high pressure applications. As figure 2 illustrates, it consists of an outer shell with a bundle of tubes inside it, the tubes are either running straight or in a “U” shape. One fluid runs through the tubes, and another fluid flows through the shell surrounding the tubes to transfer heat between the two fluids. The set of tubes is known as the tube bundle.
Heat is transferred from one fluid to the other through the walls of the tubes, either from the the tubes fluid to the shell fluid or the other way. The fluids can be either liquids or gases on either the shell or the tube. In order to transfer heat efficiently, many tubes are used in order to increase the heat transfer surface area between the two fluids.
In order to develop a comprehensive control strategy for any control loop, it’s important to identify the process variable of interest or the “controlled variable”, the manipulated variable, and the different disturbance variables that are directly affecting the controlled variable.
For the sake of this illustration, consider the heat exchanger shown in figure 3, the shell side fluid is the process fluid that is required to be heated to a certain temperature set-point, this result temperature is measured at outlet of the heat exchanger T1-OUT (controlled variable).
Heating is achieved by passing steam in the tube side, the more steam passing through the tubes the more heat is transferred to the process fluid, and vice versa. Control of the steam flow F2 (manipulated variable) is achieved by throttling a modulating valve installed on the steam inlet side.
Three major disturbances can affect the process fluid outlet temperature:
- Changes in process fluid flow rate, F1
- Changes in process fluid inlet temperature, T1-IN
- Changes in steam pressure, causing a change in steam flow rate, F2
The control objective is to maintain process fluid outlet temperature T1-OUT at the chosen set point, despite of any disturbances, by manipulating the steam flow rate F2.
In the feedback control scheme, the process variable, T1-OUT, is measured, and applied to a PID (Proportional – Integral – Derivative) based feedback temperature controller (fbTC), which compares the process variable with the desired temperature set-point, and in turn calculates and generates the control action required, either to open the steam control valve more or less.
The most important, and incomparable, advantage of the feedback control scheme, is that no matter what is the source of disturbance, known or unknown, a corrective action will be taken by the controller. Next to that is the fact that feedback control requires very little knowledge of the process, a process model is not necessary to setup and tune the feedback scheme, though it would be an advantage.
On the negative side, the major disadvantage of feedback control is it’s incapability to respond to disturbances, even major ones, until the controlled variable is already affected. Also, if too many disturbances occur with significant magnitude, they can take process out of control.
In Cascade control scheme, instead of feeding the output of the PID temperature controller to the control valve directly, it is fed as a set-point to a feedback PID based steam flow controller (fbFC). This second loop is responsible for making sure that the flow rate of the steam doesn’t change due to any uncontrollable factor (i.e. steam pressure changes or valve problems).
In order to understand how this works, consider the heat exchanger is in steady state operation and the outlet temperature is matching the set-point, the controller output of fbTC is constant, a sudden increase in steam pressure will cause steam flow rate F2 to ramp up. This will cause a change in the controlled variable after sometime, not immediately. Without the flow control loop, fbTC alone will not take any corrective action until the outlet temperature is already affected.
By implementing the cascade strategy, the feedback flow control loop “fbFC” will adjust the valve position immediately once the steam flow rate has changed to bring the flow back to it’s value of the previous steady state condition (because the flow set-point given by the temperature controller didn’t change as the outlet temperature didn’t change yet), preventing a change in the outlet temperature before it happens.
It is worth mentioning that the flow control loop must be tuned to run much faster than the temperature control loop, so it would cancel the effect of flow variance before it affects the process fluid outlet temperature.
Unlike feedback control, feedforward control takes a corrective action on the same moment as a disturbance occurs, feedforward control doesn’t see the process variable, yet it only sees the disturbances and responds to them as they take place. This enables a feedforward controller to quickly and directly compensate for the effect of a disturbance.
To implement feedforward control, an understanding of the process model and the direct relation between disturbances and the process variables is a must. For heat exchanger, a derivation from the steady state model will lead to the following equation that determines the amount of steam flow required:
F2sp = F1 × (T1-OUTsp – T1-IN) × (Cp / ΔH)
- F2sp = Steam flow rate calculated set-point, to be applied to fbFC
- F1 = Process Fluid Flow Rate, measured disturbance
- T1-OUTsp = Process fluid temperature set-point at the heat exchanger outlet
- T1-IN = Process fluid inlet temperature, measured disturbance
- Cp = Process fluid specific heat, known
- ΔH = Latent heat of vaporization for steam, known
Applying the above equation to calculate the required steam flow rate is sufficient to cancel the effects of changes of the process fluid flow rate and temperature, in a perfect world with few enhancements to the process model, this feedforward controller is enough to perfectly control the process, unfortunately, it’s not a perfect world.
The obvious advantage of this scheme is that it takes the corrective action before the process is upset. On the down side, it mandates a high initial capital cost as every disturbance must be measured, increasing the number of instruments and the associated engineering cost, also this approach requires deeper knowledge of the process, and practically, it’s not realistic to depend only on feedforward control without taking into account the measured process variable.
An integrated approach that uses feedback, feedforward and cascade control, shown in figure 7, is more than capable of answering to heat exchangers control requirements:
- Feedforward loop will handle major disturbances in the process fluid
- Cascaded flow control loop will handle issues related to steam pressure and valve problems
- Feedback loop will handle everything else
Combining the three schemes of control described above to optimize the heat exchanger operation is not only necessary to minimize process variance and maximize the in-spec product, it is also extremely important for energy efficiency requirements in petrochemical industries.