BTL Mark: Resolve interoperability issues & increase buyer confidence
PART 2: AN INTRODUCTION TO RELATIONAL CONTROL
Hartman, P E
A Three Part Series
PART 1: Why PID Control is Outdated for Modern Building Applications
PART 3: Process Change Required For Effective Relational Control
In last month’s article, I discussed performance and energy efficiency shortcomings of PID control. Back when I took control theory, I was taught that the classical feedback control - the basis of modern PID control - constituted the entire universe of real-time modulating control. The premise was that all means of control could be distilled into the classical feedback control model. That thinking may be a reason that the industry continues to see the remedy to the performance shortcoming of PID controls in expanding and furthering the development of that technology. Thus, the emphasis in the industry today appears to be focused on improving self-tuning and setpoint reset functions in an attempt to better optimize and coordinate systems by continuously adjusting and more precisely maintaining PID temperature and pressure setpoints.
TYPES OF RELATIONAL CONTROL
Newer control theory demonstrates clearly that neither the premise nor the resulting emphasis on enhancing PID control is valid. We now know that the universe of potential control approaches is essentially endless and classical feedback control occupies only a very small domain. An important element of the emerging control theory that assisted in developing this updated view is The Equal Marginal Performance Principle. Every HVAC engineer should work to become thoroughly acquainted with the new perspective on configuring and operating systems composed of multiple power consuming components that this powerful principle provides. The development of the Equal Marginal Performance Principle offers an enlightened vision of system-wide energy optimization with the realization that far more effective, simpler and more direct energy optimization relationships exist than what are required when one attempts to optimize by adjusting temperature and pressure setpoints. Demand based control has been developed to implement this simple, direct optimized control. Demand based control is one type of relational control that operates components in a system based on the real-time power consumption of each component in a multiple component system.
A second type of relational control is based on the iterative logic that has long been used to solve complex mathematical problems. Iterative logic is also very well suited to provide modulation and capacity control. With modern digital controls, the iterative process is easily applied to control applications. This type of relational control is effective when a single controlled device or function must satisfy a number of clients, or when the operation of such a device depends on a number of factors. Examples include the control of a fan that serves multiple VAV boxes, or the control of a non-linear modulating hydronic cooling coil valve or a VAV box damper.
A comparison of various features of these two types of relational control with those of classical feedback control is shown in Table 1 below.
Table 1 compares some of the features of the two types of relational control described above with features incorporated in classical feedback control that is the basis of the PID control widely employed today. It is important to note that relational control is by no means limited to the two types displayed in this chart. Rather, the opportunity for development of new types of relational control is virtually unlimited. I expect that dozens or perhaps even hundreds of new types of relational control types will be developed and described over the next decade as this emerging control approach becomes more widely employed as a modern replacement for classical feedback control.
COMPARING PID AND RELATIONAL CONTROL
If one compares the features of the two types of relational control with PID control in Table 1 it is apparent there are several important differences. Note that the required response time interval for PID control is an order of magnitude shorter than what is required for relational control approaches. The short response time required by PID along with the need to evaluate precise changes in the controlled variable value in that time make it difficult for many DDC systems to employ their network to operate PID control effectively. The use of modern DDC networks is thus discouraged for control and remote sensors used in PID control loops are typically wired all the way back to the control panel that operates the controlled device rather than connected to a more convenient point on the DDC network.
Perhaps the most important difference is energy optimization. Energy optimization with PID control requires an entirely separate step. The only way it can be accomplished is by continually adjusting the various controlled setpoints. But continually changing setpoints leads to additional stability issues and is a reason operators often discontinue setpoint reset functions. Finally, the basis of control for the classical feedback control is a proportional relationship between a change in controller output and controlled variable response. Over the wide ranges of operating conditions that are required for optimum control, this proportional relationship is very difficult to achieve. Configuring HVAC systems in an attempt to maintain proportional control relationships often results in unneeded design features that add cost and add to system energy use.
RELATIONAL CONTROL APPLICATIONS
Demand Based Control: One of the most dramatic applications for demand based control is in the coordination of the chiller plant and fan operation to cool building spaces. The Equal Marginal Performance Principle states that such a system is optimized only if the same added power to the chiller plant or the fan system(s) will result in the same additional cooling effect to the spaces served. Current practice is to distribute chilled water at one temperature and provide supply air at another temperature. As demand for cooling increases, independent PID loops provide additional airflow, and adjust valves and plant operation to maintain temperature setpoints. Additional power is supplied to fan and the chiller plant as the load increases. But the question is: “How can we operate the system to ensure that the ratio of power added to the chiller plant and fan system(s) results in optimized overall system operation?”
This is a very difficult task to accomplish with conventional temperature setpoint control. As practiced today, such optimization requires the continuous reset of chilled water and supply air temperature setpoints based either on a complex array of real-time information or on a reset table. However, the old adage applies – you can’t control what you don’t measure. Since neither of these optimization techniques actually measure power use, neither accurately results in effective energy optimization.
However, with demand based control, a simple control algorithm monitors actual power consumption of the chiller plant components and fan unit(s) in response to changes in space cooling loads. Under demand based control, neither the chilled water nor the supply air is maintained at fixed temperature setpoints. Instead, the demand based control algorithm coordinates the operation of the chiller plant, chilled water distribution system and the fan unit(s) so that as cooling load changes occur, optimized operation is maintained by regulating the increase (or decrease) in power to each component in the ratios required to meet the changing cooling load optimally.
Beyond energy optimization, demand based control has many benefits when compared to the present method of operating individual components at specific temperature or pressure setpoints. Among the most important is the elimination of capacity problems caused by low delta T syndrome and far more stable system operation. While the electric energy use with such control is often half or less of conventional PID control, the side benefits of improved capacity and operational stability are sufficient reasons on their own to adopt this new and simpler method of control.
Intelligent Iterative Control: Another type of relational control useful in HVAC control is intelligent iterative control (IIC). A good example of how intelligent iterative control can improve system operation is its application to modulate VAV box dampers. With conventional PID control, VAV box dampers are often difficult to control, particularly at low flows. The reason is that the airflow to damper position relationship for many such dampers is non-linear in the extreme. In addition, that relationship is very dependent on actual duct pressure conditions at the box which is unknown and for many systems not an entirely stable parameter itself. To maintain stable operation and to ensure damper repositions are not excessive, many conventional control techniques involve substantial control deadband, resulting in imprecise control. However, with intelligent iterative control, it is possible to control such a non-linear function with precision by considering factors beyond an error signal from an airflow setpoint.
The purpose of VAV box damper control is to distribute cooling air to multiple spaces and often to maintain a minimum quantity of air for space ventilation. Using the control system network, all the VAV boxes served by a fan system can be analyzed together to determine if more or less cooling from the system is required, and then how the distribution can be adjusted to satisfy the spaces more evenly or satisfactorily. This is not a difficult problem to solve, and by knowing the current operating conditions of the fan system, and the VAV boxes served, algorithms can be developed to control the box dampers more effectively so that they can operate at lower pressure and thus lower air energy requirements.
RELATIONAL CONTROL AND DEMAND RESPONSE
One of the most compelling reasons to accelerate the change from PID based control technologies to network based relational control is the growing understanding that substantial improvements in source energy optimization can be achieved by coordinating the end use conversion of energy with the upstream source – the electric utility generation and distribution system. At present, this is called Demand Response, and it is an increasingly popular approach to regulate electric energy use during peak use periods. Relational control, because it is network based is a perfect fit as the control vehicle to employ Demand Response. Implementing Demand Response with PID based systems can be extremely cumbersome due to the stand alone nature of the control. If one executes conventional demand limiting on the chiller plant, distribution pumps and supply fans will likely speed up as chilled water temperature rises and negate some or all of the energy reduction from the chillers. Coordinating all of the stand alone controls in a PID operated system for it to react effectively to Demand Response is a complex undertaking. But with relational control, the network coordination and optimal relationships at all loads are already incorporated in the control. It is not at all difficult to add the demand response as another control constraint and respond to Demand Response requirements effectively – and at all times optimally.
NEW VISTAS WITH RELATIONAL CONTROL
With the development of alternatives to PID control, it is becoming clear that the industry needs to make a paradigm shift in the control technologies applied to HVAC applications. Consider that a basic purpose of any HVAC system is to provide comfort for building occupants which is not a single fixed point of operation. With an understanding of relational control, it’s becoming easy to visualize how one can configure and operate HVAC systems without any PID control loops at all. In so doing, it’s also easy to see how much more effective and efficient such systems can be. Indeed, relational control provides bright new vistas for efficient and effective HVAC control.
But imagining such an improved world of HVAC control is a long way from the reality of our industry today. And it begs the question of why we have yet to really embark on such a shift in technologies. In my view the answer lies in large part with the crude and convoluted process our industry employs to conceive, implement and support control technologies embedded in HVAC systems. In the final part of this series, I will discuss some industry structural changes that need to be considered if we are to be successful in reaching the brilliant future relational control allows us now to see in the distance.
[Click Banner To Learn More]
[Home Page] [The Automator] [About] [Subscribe ] [Contact Us]