Considerations for Regenerative Combustion Systems
Controls strategies, media selection and cycle time affect combustion system performance for industrial heating.
The frequent cycling of a regenerative combustion system introduces nonlinearities and discontinuities that place challenges on a traditional combustion controls system. Several approaches can minimize these effects. This article will discuss some of these concepts, introducing strategies to simplify and improve control of a regenerative system. It also will describe factors that influence system design.
Regeneration is a common form of heat recovery originally developed in the 1850s — and vastly improved since then. It can yield combustion efficiencies as high as 75 percent on a gross heating value basis for a furnace temperature of 2000°F (1000°C). By contrast, for conventional cold-air firing, 30 to 35 percent combustion efficiency is common. With regenerative combustion systems, fuel usage for a given process can be less than required for other methods.
Regenerative burners work in pairs: One burner fires while the other recovers heat from the waste gas by passing it through a porous bed of ceramic or metallic media. As the waste gas passes through the bed, it gives up heat to the cold media. In this way, the waste-gas temperature falls to 300 to 400°F (150 to 200°C). After the media bed has absorbed the heat, the cycle reverses, and cold air enters the now-firing burner. As the cold combustion air passes through the media, the heat transfer reverses, and the combustion-air temperature increases to within 200°F (110°C) of the chamber temperature. The cycling occurs at regular intervals with each burner — in turn — firing and exhausting. Various combinations of cycling valves for air, fuel and exhaust (waste gas) control the flows to each burner for firing and exhausting.
Loop Control for Regenerative Burners
In general terms, a control system tries to match a measured value — also called a process variable (PV) — to a target setpoint value (SP) by manipulating a control variable (CV).
For example, assume a process needs a specific process temperature (TP). A gas valve in the system manipulates the control variable (CV) in response to the difference — known as ΔT — between the specific process temperature (TP) and the actual temperature (TA). In practice, the system measures the actual temperature (TA). If there is a positive difference (ΔT), then the system signals the gas valve to open, thus increasing the heat. Opening the valve at the proper speed to ensure an appropriate response is tricky. If it opens too fast, then the temperature will overshoot. If the valve opens too slowly, the slow response will make the system sluggish.
Historically, control of a regenerative system is accomplished with proportional-integral-derivative (PID) loops. Each controlled variable — among them, temperature, airflow, gas flow, furnace pressure control and ratio control — has a separate control loop. The response of these loops depends on three factors: the proportional (P), integral (I) and derivative (D) parameters. Because each valve responds in a unique fashion, the values of the parameters generally are different for each control loop. Making the process of setting up such a control system even more complicated, each of these loops depends on the others, either directly or indirectly. Therefore, the process of selecting the proper P, I and D requires sometimes extensive tuning, and it often is time consuming.
In effect, with conventional PID control, the control system looks at the ΔT and makes a binary choice. For instance, is ΔT equal to 0? If yes, the control does nothing. If no, the control effects a change to the system.
As an alternative, an approach known as fuzzy logic can be a good application for regenerative control. As opposed to a binary choice, fuzzy logic looks at the ΔT and assigns a response relative to its value. In other words, under fuzzy logic, the response of the system depends on the relative difference between the setpoint (SP) and process variable (PV), and it weighs other related factors as well.
As an example, take the furnace pressure control loop. Under a PID control scheme, the furnace pressure output is controlled directly to the output of the PID loop with a pressure transmitter as a PV. Under a fuzzy logic control scheme, the logic examines a combination of other inputs — any changes in firing rates, or how recently the door has opened or closed, for example — and gives a weight to each of these factors when calculating whether the furnace pressure is acceptable. By factoring in the relative importance of these events, the system can decide if or how to change the damper position. It will move the damper by an incremental amount before checking all of these factors again. This approach prevents hunting for the correct position and avoids oscillations.
Implementation of fuzzy logic for control is more complicated than this example might indicate, especially when loops begin to interact. However, fuzzy logic has shown promise for regenerative systems because it can handle nonlinearity as well as step-wise discontinuities better than conventional controls can.
Cycle Time Affects Control and Burner Selection
The length of time between cycles has an impact on control and burner selection. During the switch-over, there are a few seconds of dead time when neither burner is firing, so the system must deliver the necessary heat in a shorter time than it would seem. For example, if the cycle time is 40 seconds, then the system must deliver the required average heat in 38 (40-2) seconds, meaning that the actual capacity of the burner must be about 5 percent higher than the average input.
Figure 1 shows the relation of cycle time versus increase in capacity. The actual capacity of the burner becomes larger as cycle times become shorter. For example, if a process requires 10.0 MM BTU/hr of power, then for a 20-second cycle, the actual burner capacity would be:
1.11 x 10 = 11.1 MM BTU/hr
For a 90-second cycle, the power is 10.2 MM BTU/hr. Figure 1 shows the relationship between cycle time and required actual capacity.
Longer cycles typically require more media in order to ensure that there is enough thermal capacity in the media to hold all of the heat. However, the efficiency suffers with the length of each cycle.
Figure 2 shows a qualitative comparison of exhaust temperature and air preheat as cycle time increases. For very short cycles, there is rapid transfer of energy from the exhaust gas to the combustion air, with very high preheats. In the extreme case of an infinite cycle time, the burner acts like a cold-air burner — with the resultant low combustion efficiency. Selection of a proper cycle time generally requires balancing these many considerations.
Media Options for Regenerative Combustion Systems
Proper selection of the media used can have an impact on maintenance and performance as well as capital costs for a system. The two major types of media are cake (also known as honeycomb) media and ball-type media. While both types of media perform the same basic function of storing and giving up heat in sequence, the specific application warrants careful selection for proper performance.
Honeycomb or cake-style media is a matrix of some kind of ceramic with interconnected air-filled voids. It resembles, in many ways, a sponge. By contrast, ball-type media are small spheres of a ceramic material. Because spheres do not pack tightly when piled in a media case, there are voids that allow flow of air or exhaust gas.
There are applications for either type of media. Cake-type media generally is lighter and requires less volume for the same heat transfer, but it is susceptible to clogging due to any impurities in the waste-gas stream. It is not easy to clean cake media if it becomes fouled. The ball media is physically heavier, but it is more forgiving of impurities in the waste gas. It is also relatively easier to clean than the cake media.
In the case of the media balls, it also is important to consider their size. Typical media ball sizes range from roughly 0.375 to 0.75”. The smaller media balls do a better job of transferring heat — due to the fact that the surface-area-to-volume ratio is higher — compared to the larger balls. However, because they pack more tightly, plugging is more common, and the pressure drop across the bed is larger. This may necessitate larger combustion air and exhaust fans. As a consequence of the smaller media, the spaces between the balls are smaller, and thus the media depth (and the media case size) is less for smaller media balls (figure 3).
In conclusion, while regenerative technology has been around for many decades, continuous improvements in materials and control require proper selection for proper performance. Factors such as controls strategies, media selection and proper choice of cycle times affect combustion system design as well as overall system performance.