PREDICTIVE CONTROL OF A DX A/C SYSTEM FOR ENERGY EFFICIENCY

Get Complete Project Material File(s) Now! »

BACKGROUND

The total energy consumption of the world market will increase by 36% between 2008 and 2035 [1]. This energy consumption can be divided into four main sectors: residential, commercial, transportation and industrial. The rise in energy consumption by buildings has been significant in most part of the world. The building sector accounted for 21% of energy consumption in the world in 20081 . In the USA, the building sector accounted for 41% of primary energy consumption among all sectors in 20102 . In South Africa, the residential, commercial and public services sectors share 40% of electricity consumption, and based on National Electrification Statistics, electricity consumption is expected to increase constantly3 . Because of the increase in the population, enhancement of comfort, global climate change and more time spent indoors, energy consumption in the building sector displays an upward trend.
Therefore, most countries are focusing on the building sector as having the greatest potential for energy savings necessitated by increasing energy demands, energy price and environmental issues. Building’s energy demand is the main reason of electricity consumption owing to rapidly escalat- ing space environmental quality requirements (such as thermal comfort, indoor air quality (IAQ), ventilation, refrigeration and so on), which leads to significant greenhouse gas emissions. Energy consumption demands in buildings are directly related to air conditioning (A/C) systems. A/C systems account for half of a building’s energy usage. At the same time, owing to greenhouse effects the global temperature is increasing steadily, which in turn has caused an increase in the use of A/C systems. Accordingly, lowering the energy consumption in A/C systems of buildings is an important factor in lowering greenhouse gas emissions. The study reported in this dissertation is motivated by the considerable energy efficiency and energy cost potential in the building sectors. Energy management of building A/C systems has become important and hence the need to lower the energy cost and improve the energy efficiency of buildings. In addition to improving the energy efficiency of buildings, interventions can be made in other ways [2, 3], such as appliances operation scheduling [4, 5, 6], hybrid energy system supplies [7, 8, 9], generator maintenance scheduling [10], lighting retrofitting and metering schemes [11, 12, 13, 14], envelope and whole building retrofitting schemes [15, 16], facility retrofitting and maintenance schemes [17, 18, 19] and an energy-water nexus [20, 21]. Therefore, building energy efficiency is a very broad field involving multiple layers and focuses. Energy efficiency improvement can also be effected in other sectors such as transportation scheduling including, overhead cranes [22] and belt conveyor [23] and industry application aspects [24, 25]. The main topic of the thesis is the scope of the energy management strategies of building A/C systems, which will be introduced in the following sections.

Indoor comfort control

Lowering building energy usage should not sacrifice user benefits [26]. User-health is related to energy efficiency in some aspects. Nowadays, more and more people are working indoors much of the time, therefore providing high thermal comfort and IAQ levels for users would contribute to increased work efficiency and productivity. Effective energy management of building A/C systems can ensure IAQ, thermal comfort and energy efficiency. In [27], the authors designed two controllers for a heating, ventilation and air conditioning (HVAC) system, including feedback and feed-forward controllers, to improve the control performance (smaller temperature variations) as well as energy efficiency. The use of optimization algorithms in various applications related to energy management in building A/C systems has been increasing significantly over recent years. A multi-objective genetic algorithm was used [28] to obtain predictive control of air conditioned systems. The simulation results demonstrated that this method would achieve good temperature regulation with important energy savings. Ensuring indoor air humidity at an acceptable level is an important factor since it has a direct impact on indoor thermal comfort level and the operational efficiency of buildings’ A/C installations [29]. In fact, in cities with highly humid climates, such as Cape Town or Hongkong, high humidity may still adversely affect indoor thermal comfort level and the energy efficiency of building A/C systems, even when indoor air temperature has been maintained at a desired value.
Various humidity control approaches have been applied in large-scale central and chilled water A/C systems, such as pre-conditioning outdoor air and heat pipe technologies [30, 31], or chemical and mechanical dehumidification desiccant mechanisms [32, 33, 26, 34]. A wavelet-based artificial neural network (ANN) with a proportional- derivative (PD) controller was proposed [35] for an A/C system to control indoor relative humidity and air temperature, where thermal comfort and energy efficiency of the system was improved. For controlling indoor humidity, a chiller/boiler was added to this A/C system. For controlling indoor relative humidity and air temperature, an evaporation pressure control method was addressed [36] based on the evaporator pressure and the relative humidity readings. Nowadays, IAQ has become increasingly important and is regulated by A/C system design and control. The indoor relative humidity, air temperature and carbon dioxide (CO2) concentration have been regarded as the three major factors of indoor thermal comfort and air quality.
In recent years, more and more researchers have focused on how to improve thermal comfort and IAQ as well as energy efficiency. In [37], a genetic algorithm was used to find the optimal settings of the multiple variable process (i.e. air handling unit supply air temperature, outdoor ventilation rate and chilled water temperature setpoints) by optimising the cost function including thermal comfort, energy consumption, IAQ, maximum allowed relative humidity and minimum allowed ventilation flow to reduce energy usage while keeping multi-zones’ thermal comfort and IAQ at acceptable levels. A multi-objective optimization was proposed in [38] to optimize indoor air condition for HVAC system in order to achieve high thermal comfort and acceptable air quality for occupants with efficient energy consumption all the time. However, they did not consider the coupling effects between indoor CO2 concentration, relative humidity and air temperature. In many cases, these coupling effects cannot be ignored. In fact, the experimental results [39] illustrated that the indoor CO2 concentration correlated with indoor air temperature. Furthermore, the indoor CO2 concentration is affected by air humidity as discovered through measurement investigation and data analysis [40]. As far as the researcher knows, there are few studies in the literature that discuss how to control indoor air temperature, relative humidity and CO2 concentration simultaneously, taking into account the coupling effects between them and the energy efficiency of A/C systems.

Peak demand control

In addition to reducing overall energy consumption by building A/C systems, another significant need for building controls is to reduce peak power demands. Because buildings, especially commercial and office buildings, mainly consume energy during peak hours, the peak-average ratio (PAR) can be high in the electricity grid [41]. Both electricity suppliers and customers are extremely focused on peak demand because of economic and environmental challenges. New power plants are constantly being built every year; however, this still merely serves to remit the rapidly increasing peak demand, which reduces efficiency in non-peak hours and leads to higher energy costs for buildings. Therefore, it is of great importance to use advanced strategies to reduce or shift the peak demand. Demand response (DR) as a concept has paid more attention to buildings as an effective way to reduce the peak demand.
DR encourages end-users to adjust their electric usage in good time based on electricity price. Advanced electricity rate structures including real-time pricing (RTP), time-of-use (TOU) and critical-peak-pricing (CPP), are usually applied by utilities. It is illustrated that with these time-varying rate structures users have opted to reduce their energy cost by taking DR action [42, 43, 44]. In [45], a cost-effective control scheme was proposed for building HVAC system to shift the energy usage away from the peak hours while thermal comfort and IAQ levels are maintained. For reducing peak demand, most buildings are operated according to a simple on-off strategy: the A/C systems are turned on during the occupied period and turned off otherwise. The setpoints of thermostats are usually fixed during the entire occupied period. This strategy is simple but not optimal for energy efficiency or cost-effectiveness [46]. Some alternative strategies are proposed to reduce significant peak demand by adopting pre-cooling (or pre-heating) during a non-occupied period and changing setpoints during peak hours. These strategies use the building’s thermal mass to shift the power demand. However, the setpoint schedule is pre-determined, which does not consider the time-varying outdoor and indoor conditions and the states of HVAC systems. A number of researchers have focused on reducing the peak demand by adjusting the operational scheduling of HVAC systems [47, 48]. However, these methods are not able to deal with the impact of disturbances such as buildings’ internal loads and weather conditions. Therefore, it is of great interest and potential to develop advanced approaches to energy and cost savings for building A/C systems as well as handling disturbances.

Model predictive control

Model predictive control (MPC) has become one of the most successful advanced control strategies, which is capable of maintaining a comfortable temperature and achieving energy efficiency in buildings [49, 50, 51]. Other advantages of an MPC algorithm for building HVAC systems include robustness, tunability and flexibility [52]. The MPC strategy has also been applied in intervention strategies for other sectors, such as human immunodeficiency virus (HIV) infection [53], the dense medium coal washing process [54], power dispatch problems [55] and transportation systems [56]. Therefore, it has inspired many researchers employing MPC algorithms for HVAC systems to enhance both indoor thermal comfort and energy efficiency [57, 58, 59].
An MPC strategy is capable of improving potential building energy efficiency and thermal comfort and the performance is better than conventional PI controllers reported in [49]. In [60], an economic MPC method was presented for optimizing the building demand and energy cost under a TOU price policy to improve temperature comfort and reduce demand, as in [61, 62]. In [60], the simulation results showed that the MPC strategy is able to shift the peak demand to off-peak hours and reduce energy costs more in comparison with a baseline case. However, the goodness-of-fit for temperature and power models is 76.22% and 80.3%, respectively, which mean that the non-accuracy is approximately 24% and 20%, respectively, due to modelling a linear system. The authors have also tested on particular days to show the advantage of MPC over its baseline strategy, but it can be seen from the tests that load shifting on particular days is no more than 24%, which implies that the performance of the MPC strategy was unsatisfactory. In [63], the authors proposed a hybrid model predictive control (HMPC) scheme, including a classical MPC and a neural network feedback linearization, to minimise the energy and cost of running HVAC systems in commercial buildings.
Though the study modelled a nonlinear HVAC system, it was to be controlled based on a linearized system by feedback linearization, which added control variables and in turn increased computational complexity. This control method was also used in [64]. In [65], an MPC strategy was presented to reduce energy cost and demand while keeping indoor temperature at a comfort level. To facilitate the MPC controller, the nonlinear HVAC system is linearized around an equilibrium point. This equilibrium point is obtained by fixing the supply fan and solving a state point. In [66], the authors proposed a practical cost and energy-efficient MPC strategy for HVAC load control under dynamic real-time electricity pricing. The simulation results displayed that the MPC strategy can lead to significant reductions in energy consumption and cost for occupancies. However, the building’s nonlinear thermal model is linearized and used as the plant to be controlled by the presented MPC strategy. A model-based predictive controller combined with a building energy management system was purposed to improve indoor environmental quality, including CO2 concentration, air temperature, relative humidity and illuminance, and minimize energy costs in [67]. However, the study did not consider the coupling effects between them.

TABLE OF CONTENTS :

• CHAPTER 1 INTRODUCTION
  • 1.1 BACKGROUND
  • 1.2 A/C CONTROL SYSTEM FRAMEWORK
    • 1.2.1 Indoor comfort control
    • 1.2.2 Peak demand control
    • 1.2.3 Model predictive control
  • 1.2.4 Hierarchical control
    • 1.2.5 Large-scale A/C system control
  • 1.3 DX A/C CONTROL SYSTEM FRAMEWORK
    • 1.3.1 DX A/C system
    • 1.3.2 DX A/C system control based on empirical-based models
    • 1.3.3 DX A/C system control based on physical-based models
  • 1.4 RESEARCH GAP
  • 1.5 SCOPE AND OBJECTIVES
  • 1.6 RESEARCH CONTRIBUTION AND LAYOUT OF THE DISSERTATION
• CHAPTER 2 PREDICTIVE CONTROL OF A DX A/C SYSTEM FOR ENERGY EFFICIENCY
  • 2.1 INTRODUCTION
  • 2.2 CHAPTER OVERVIEW
  • 2.3 DX A/C SYSTEM DESCRIPTION
    • 2.3.1 DX A/C system
    • 2.3.2 A/C space
    • 2.3.3 Single-zone DX A/C model
  • 2.4 ENERGY-OPTIMISED OPEN LOOP OPTIMISATION
    • 2.4.1 Open loop optimal controller
    • 2.4.2 Linearization for closed-loop control
  • 2.5 CLOSED-LOOP CONTROL
    • 2.5.1 Discrete-time linear system
    • 2.5.2 Cost function
    • 2.5.3 Constraints
    • 2.5.4 The proposed MPC algorithm
    • 2.5.5 Features of the proposed MPC strategy
  • 2.6 RESULTS AND DISCUSSION
    • 2.6.1 MPC with open loop optimal controller
    • 2.6.2 Analysis of energy efficiency
  • 2.7 CONCLUSION
• CHAPTER 3 AUTONOMOUS HIERARCHICAL CONTROL OF A DX A/C SYS TEM
  • 3.1 INTRODUCTION
  • 3.2 CHAPTER OVERVIEW
  • 3.3 MODIFIED DX A/C SYSTEM
    • 3.3.1 The DX A/C model
    • 3.3.2 Load models
    • 3.3.3 PMV index
    • 3.3.4 Energy models
    • 3.3.5 System constraints
    • 3.3.6 TOU strategy
  • 3.4 HIERARCHICAL CONTROL DESIGN
    • 3.4.1 Upper layer
    • 3.4.2 Lower layer
    • 3.4.3 Algorithm
  • 3.5 SIMULATION AND RESULTS
    • 3.5.1 System setup
    • 3.5.2 Two control strategies
    • 3.5.3 Comparison of two control strategies
    • 3.5.4 Sensitivity analysis
  • 3.6 CONCLUSION
• CHAPTER 4 DISTRIBUTED CONTROL OF AN ME A/C SYSTEM
  • 4.1 INTRODUCTION
  • 4.2 CHAPTER OVERVIEW
  • 4.3 ME A/C SYSTEM
    • 4.3.1 Description of an ME A/C system
    • 4.3.2 Dynamic model of an ME A/C system
    • 4.3.3 Simplified energy models of an ME A/C system
    • 4.3.4 PMV index
    • 4.3.5 Constraints
  • 4.4 CONTROLLER DESIGN
    • 4.4.1 Upper level: Steady state optimization problem
    • 4.4.2 Lower level: DMPC
    • 4.4.3 Algorithm
  • 4.5 CASE STUDY
    • 4.5.1 Parameter selection
    • 4.5.2 Comparison of optimal scheduling control strategies
    • 4.5.3 Simulation test for proposed control strategy
  • 4.6 CONCLUSION
• CHAPTER 5 CONCLUSIONS
  • 5.1 FINDINGS
  • 5.2 FUTURE WORK
  • REFERENCES
  • ADDENDUM A SYSTEM MATRICES
  • A.1 System matrices

GET THE COMPLETE PROJECT