Photo Credit: www.cedengineering.com

Definition of a ‘good’ air conditioning in a workspace or an environment is not always clear. There are standards that codify the temperature, air flow rates, humidity etc based on the application and usage of the space that is being air conditioned, however, does the occupant always feel that the temperature is just right? The standards themselves have evolved over a period of time to factor in new technologies, new environments and also the expectations of the occupants of the workspace, be it a modern office, the mall that everyone goes on the weekend, the data center or the hi-tech factory manufacturing the handheld devices. Thus, for a chiller plant manufacturer, AHU designer, ventilation expert etc, the task of developing HVAC systems that can adapt to the requirements of users is very critical as the requirements do not remain static.

The key output for a HVAC system is occupant comfort (for spaces where humans are present). Irrespective of the external and internal conditions surrounding the occupants, the output remains the same – it may vary in amplitude (hotter or colder, humidity levels etc.) but the users of the space expect that they are comfortable and can perform the task they are present for (e.g. in a movie hall, the temperature should be neither too cold nor too hot for the audience so that they can enjoy the latest Star Wars offering without worrying about the temperature inside the movie theater). The inputs to the HVAC system are, however, many – outside temperature, internal varying load, condition of the equipment such as cooling towers, condenser tubes etc., individual preferences etc. Thus, the control of the HVAC system becomes complex and complicated. However, an effective control system can match the occupant’s requirements with the operating costs and energy usage and thus, allow the O&M team to optimise the usage of the HVAC system. There is, thus, a continuous improvement of HVAC systems control technologies which enable the manufactures and designers to keep pace with the changing requirements.

Overview of Control System

A control system typically has a single or multiple input, the system on which the control system acts, single or multiple outputs, the controller and a feedback system for the controller. Figure 1 represents the components of a control system and how they interact.

Based on the type of control, and what is being controlled, there are various strategies of control systems in use. Different HVAC components will have different control requirements – a basement exhaust fan will have a simple on/off control depending on the level of carbon monoxide in the basement, an AHU will need a controller for the air side as well as the chilled water side and both will have to be matched which a chiller may need very advanced control systems for ensuring that the compressor operates in tandem with the load on the system as a whole. Control systems are typically divided in to open loop (no feedback system) and closed loop (where the condition of the output is used as a variable in the input to the controller). The types of control approaches in these two types of systems are as follows:

Open Loop Systems

On – Off controllers: These are simple systems which are not hardwired to the state of the output. The system is simple, stable and reliable but does not cater to external changes automatically. When a window air conditioner is switched on at a particular time of the day and switched off at another without any linkage to the ambient temperature, the system is an example of a “on – off” control system. Similarly, a room heater which switches on and off based on the thermostat setting is using an on – off control approach. These systems, while easy to implement are not very accurate and reliable and used only for simple systems and requirements. Figure 2 shows a simple open loop control system.

Closed Loop Systems

In a closed loop control system, the output is maintained at a desired level by giving the state of the output as a variable to the input of the controller as shown in figure 3. Typically, a system will remain in its state of equilibrium (operating condition) till there is an external disturbance which move the system to a different operating point. When one is travelling in a car on a flat road with the accelerator pressed half way down to maintain a constant speed, if there is an uphill section, the speed of the car reduces unless the accelerator is pressed further. This is an example of a manual closed loop control system where an incline (disturbance) causes a change in the output (speed) which results in a change in the input (how much the accelerator is pressed). This then results in the output.

There are a number of different ways a closed loop control system can function. The key attributes of a control system are as below. The control system parameters are assessed based on the time and amplitude of the response and represented according on graphs.

 Rise time: The time the system takes to move from its steady state to a per cent of the final desired state.
 Steady state error: The error in the system once the control action has been implemented and equilibrium has been achieved
 Damping: The amount of control that the system provides to achieve a desired output and prevent oscillation between the set point
 Settling time: The time the system takes to reach a per cent of the final output

Types of Closed Loop Controls: The main types of control mechanisms are:

Proportional Control (P): In controllers using simple proportional control, the output is proportional to error rate. Error rate is the difference between the actual system output vs the desired output. These controllers are easy to set up and ‘tune’ but the system always has an ‘offset’ due to the nature of the control action. An example of a proportional control would be the chilled water flow control valve of an AHU. When the chilled water returns, temperature increases (a higher deviation from the desired output, or an increase in the error rate), the valve opens to allow more chilled water to enter the AHU coils. This leads to a lowering of the chilled return water temp and the valve settles to the new position. Proportional controllers are typically used where the system changes occur over a long period of time, such as the internal ambient temperature in a work space.

Proportional Integral Deferential Control (PID): These controllers, the rate of correction of the error increases as the magnitude of the output deviation increases. Thus, in the above example of the AHU chilled water valve, when the return temperature is very far from the set point, the rate of opening of the valve is faster whereas when the temperature difference is less, the valve opens slowly. These controllers allow the system to have a smaller offset and a faster response to changes. Due to the increased complexity of the action, the system is relatively more expensive and difficult to tune.

Advanced Control Systems for HVAC Applications – Model Predictive Control (MPC)

With the advances in technology and also metallurgy, the capacities of chiller plants have increased rapidly and the applications are varied as well. Traditional control systems are then not responsive enough of reliable enough to cater to the rapidly changing requirements of the workspace. Control systems have also evolved as have the chiller and HVAC system design. Some of the key trends in the HVAC chiller control methodology are
as below:

Model Predictive Control (MPC): The traditional PID controllers or the on – off controls used in building HVAC systems provide localised control and are not integrated with the whole building as the number of variable are far too many. Thus, there are advanced PID controllers at the work space level, or at the AHU level of the chiller level, but none for the whole building system. This is where the MPC control approach works more effectively. Humans predict the future scenarios of their approach and the control strategy they adopt is based on their prediction of how the future will look like. For example, when a person makes food, he or she predicts how hot the surface will become and accordingly put in the ingredients or lower the flame to match the requirement of the material that is being cooked. Similarly, in HVAC systems, the future state, based on pre-defined constraints as well as inputs is analysed at every step of the control process and the system makes corrections based on the analysis.

A model of the HVAC and building system is developed and controlled as well as uncontrolled inputs are defined, based on which, when the model is run, predictive outputs are arrived at. These predictions are used on the actual system to arrive at the best possible changes to be made to the system to meet the desired outputs. The MPC approach, since it allows the system to operate closer to the design point by handling constraints allows a more efficient use of the system resulting in lower operating costs. The MPC approach is a feed forward system, and a more refined version of the open loop system, the main difference being that in MPC, after every step of control, the future condition is predicted and corrections made. Figure 4 shows the working of an MPC system. MPC is now gaining traction in whole building design and control thanks to the availability of superior computing power at the controller level as well as on the cloud. A large amount of calculations is needed to make the MPC successful and the modern computing ability of systems allows for that.

Conclusion

The traditional approach of implementing control strategies for individual complaints and then matching them to the other components is now no longer effective due to the higher demands of the building environment as well as fast changing requirements. In addition, the cost to operate HVAC systems is increasing and new control strategies are needed to optimise cost. Advanced, but localised PID controllers are now giving MPC systems. These controllers have computational capabilities which allows the device to predict the future output based on the current state, and where required, made necessary changes to achieve the desired set point. The key advantage of predictive control is that it mimics the human approach to control – decide the control strategy based on how the future is looking. As building owners and designers of HVAC systems get more aquatinted with this approach, there will be more installations of HVAC systems with MPC as the backbone.