Control strategies for integration of thermal energy storage into buildings: State-of-the-art review

Accepted Manuscript Title: Control strategies for integration of thermal energy storage into buildings: State-of-the-Art review Author: Zhun Jerry Yu Gongsheng Huang Fariborz Haghighat Hongqiang Li Guoqiang Zhang PII: DOI: Reference:

S0378-7788(15)30011-6 http://dx.doi.org/doi:10.1016/j.enbuild.2015.05.038 ENB 5886

To appear in:

ENB

Received date: Revised date: Accepted date:

13-1-2015 16-4-2015 22-5-2015

Please cite this article as: Z.J. Yu, G. Huang, F. Haghighat, H. Li, G. Zhang, Control strategies for integration of thermal energy storage into buildings: State-of-the-Art review, Energy and Buildings (2015), http://dx.doi.org/10.1016/j.enbuild.2015.05.038 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Control strategies for integration of thermal energy storage into buildings: State-of-the-Art review Zhun (Jerry) Yua, Gongsheng Huangb, Fariborz Haghighatc, Hongqiang Lia, Guoqiang Zhanga a

College of Civil Engineering, Hunan University, Changsha, Hunan, 410082, PR China Department of Architecture and Civil Engineering, City University of Hong Kong, Kowloon, Hong Kong c Department of Building, Civil and Environmental Engineering, Concordia University, Montreal, Quebec, H3G 1M8, Canada

cr

ip t

b

te

Abstract

d

M

an

E-mail address: [email protected] (Zhun Jerry Yu).

us

Corresponding author. Tel.: +86 88821040; fax: +86 88821040

Thermal energy storage (TES), together with control strategies, plays an increasingly

Ac ce p

important role in expanding the use of renewables and shifting peak energy demand in buildings. Different control strategies have been developed for the integration of TES into building-related systems, mainly including building envelopes, HVAC systems and hot water tanks (HTWs). A systematic survey of control strategies using TES in the building-related systems is still lacking. This paper presents a comprehensive review of these control strategies. It provides a summary of the applied strategies and makes recommendations for future studies. Considering that control techniques serve as a basis for the implementation of control strategies, typical control techniques utilized in the building systems, as well as their strengths and weaknesses associated with the application, are also introduced to help users gain a better understanding.

Page 1 of 42

Keywords Control strategy; Thermal energy storage; Building envelope; HVAC; Hot water tank

ip t

Highlights

cr

Little study reviews control strategies for buildings integrated with TES (BITES). This study reviews typical control techniques utilized in BITES.

us

This study reviews different control strategies adopted in BITES.

Ac ce p

te

d

M

an

Further recommendations for developing more efficient strategies are made.

Page 2 of 42

1 Introduction Replacing conventional energy systems with renewables such as solar and

ip t

geothermal energy is an efficient way to reduce dependence on non-renewable energy sources as well as greenhouse gas emissions. Although considerable effort has been devoted to addressing this issue, the main drawback of renewables’ intermittence and

cr

variability in their availability remains one of the biggest barriers to building applications.

us

This drawback might result in significant mismatches between the time of building energy demand and energy production. Bridging this gap calls for effective methods of incorporating thermal energy storage (TES) into buildings. TES allows energy to be

an

stored when it is at its most abundant and to use it in time of need, and thus is well adapted for use with intermittent renewable energy sources. Similarly, by storing energy

M

during periods of low demand and using it when the demand is high, it can be used to shift energy loads from the peak periods to the off-peak periods.

d

In the past several decades, many studies have been conducted to investigate the

te

integration of TES into different building-related systems, mainly including building envelopes (e.g. wallboards, roofs and floors), HVAC systems (e.g. free cooling systems

Ac ce p

and central air-conditioning systems) and hot water tanks (HTWs). Both latent storage mediums like phase change materials (PCMs) and sensible storage mediums like building thermal masses have been used for such applications. In terms of the way to deliver thermal energy to the storage mediums, buildings integrated with TES (BITES) are generally classified into either active or passive systems. Active systems are defined as the one in which a fluid is circulated or electric heaters are used to exchange heat, while in passive systems no mechanical equipment is employed to deliver thermal energy [1-3]. For example, ice storage systems belong to active systems while building thermal masses (envelopes) are passive systems.

For BITES to reach their full potential, a key challenge to meet is the development of effective control strategies. Aiming at fulfilling building cooling and heating

Page 3 of 42

requirements, these strategies provide a flexible approach to integrating TES with other facilities, as well as responding to varying weather conditions, occupant behavior and utility rate structures. Despite playing a crucial role in system performance improvement and optimization, existing control strategies implemented in the buildings are insufficient

ip t

for deriving benefit from TES. This highlights the need to examine existent strategies and properly address the issue of developing more advanced and efficient strategies. So far

cr

limited study has reviewed control strategies adopted in BITES. For example, Sun et al. [4] reviewed peak load shifting control strategies using cold TES facilities in commercial

us

buildings. Pintaldi et al. [5] reviewed and evaluated control strategies for optimally managing TES for solar air-conditioning plants. ASHRAE [6] summarized different

an

control strategies for centralized cooling and heating systems with and without TES, focusing on strategies associated with cost optimization. However, a systematic survey of

M

control strategies for buildings integrated with TES is still lacking.

This paper presents a comprehensive review of the control strategies adopted for

d

BITES by sorting BITES into three categories: active systems, passive systems, and combined systems (active & passive). The main goal is to provide a summary of these

te

strategies and make recommendations for future studies. The target audiences are

Ac ce p

researchers, building designers and operators, system analysts and practitioners. Considering that control techniques serve as a basis for the implementation of various control strategies, typical control techniques utilized in BITES are also briefly introduced to help the target audiences gain a better understanding.

2 Control techniques utilized in BITES Control techniques utilized in BITES are divided into different categories shown in

Fig. 1 [7].

Page 4 of 42

Typical control techniques utilized in BITES

Model predictive Control (MPC)

Optimal control

Under floor heating systems withPCM

Solar power systems with TES units

Ice storage systems

Ice storage systems

Adaptive control

Neural network control

Fuzzy logic control

Still rare

Ice storage systems

Space heating systems with TES units

Reinforce -ment learning control

Hybrid control

ip t

PID control

Other control

Ice storage systems + building envelopes

an

On-off control

Soft control

Multiagent control

Reinforce -ment learning control +MPC

cr

Hard control

us

Classic local-loop control

Still rare

Ice storage systems + building envelopes

M

Fig. 1 Classification and typical applications of control techniques utilized in BITES [7] This section presents the control techniques and discusses their strengths and

te

d

weaknesses associated with the application to BITES.

Ac ce p

2.1 Classic local-loop control

Classic local-loop control techniques basically consist of on-off control and

proportional-integral-derivative (PID) feedback control. On-off control techniques are used to control variables with discrete values such as on or off. Lin et al. [8] reported an application of on-off control when incorporating PCMs into building structures. They developed an under-floor electric heating system with shape-stabilized PCM plates in order to take advantage of night electric heating (i.e., 11 PM to 8 AM). Specifically, electrical heater is used to charge the PCM during the night and then to be released to the indoor air during the daytime. They established a prototype room to investigate its thermal performance. An on-off temperature controller was used to turn on the electric heater when its temperature was below 55 °C and turn it off when its temperature was more than 70 °C. The results show that this room was thermally comfortable and energyefficient: more than half of the demand was shifted from the peak period to the off-peak

Page 5 of 42

period.

As the most ordinary but dominating type of feedback control, the principal objective of PID control is to minimize the ―tracking error‖, i.e. the difference between

ip t

the desired set-point and measured process variable.

PID control was used to control temperature or flow rate in BITES both individually

cr

and combined with other control techniques. For example, Powell and Edgar [9] simulated a TES unit used in a concentrated solar power system in order to investigate the

us

benefit of adding this unit. Two individual PID controllers were used: one to keep the output temperature of the solar power system constant and the other to keep the power

an

output of the TES constant. The simulation results show that the whole system is able to provide constant power output despite the variability and intermittence in solar radiation‘s availability. Mawire and McPherson [10] incorporated TES into a solar

M

cooking system by charging the TES with a solar collector and then discharging its heat to the cooking system. The objective of this system was to maintain a nearly constant

d

charging temperature not influenced by the variation of solar radiation, thereby efficiently transferring heat. In order to achieve this objective, they used a combined PID control

te

and internal model control (IMC) structure which consists of an outer PID control loop and an inner IMC control loop. The PID loop is utilized to remove fast changing

Ac ce p

disturbances (e.g. power from the solar collector) beyond the IMC loop‘s ability to process. The results show the efficient performance of the proposed control structure considering the charging temperature was maintained within a few degrees of the setpoint.

The advantage of PID control lies in its intuitiveness and relative simplicity [11].

Also, no precise mathematical models of the control process are necessary. However, LeBreux et al. [12] reported that in many space heating applications, the PID controller is insufficient to control solar and electric energy storage. The main reason is that it tends to result in overheating due to its weakness in dealing with energy storage‘s time evolution characteristics and in processing disturbance inputs. Moreover, PID controllers are normally used at the component level instead of the system level. It is seldom used

Page 6 of 42

independently when overall building performance is taken into consideration [13]. In addition, PID controllers need to be tuned for each single room when TES is embedded into building envelopes. Such tuning is always time-consuming and difficult.

ip t

Nowadays the majority of classic local-loop control techniques are realized in practice with a programmable logic controller (PLC) which, as programmable

cr

microprocessor-based devices, can be programmed by users to store instructions and accordingly perform a sequence of actions. For example, Farahani and Saeidi [14]

us

designed and implemented an ice storage system for an office building in order to demonstrate TES‘s cost saving benefits and promote its practical application. They

an

divided the cooling system operation into different operation modes (e.g., chiller only, ice and chiller, etc.). Based on the evaluation of various control techniques, a PLC was selected and programmed for all the operation modes' algorithms in response to demand

M

and as a function of time. The results indicated that the PLC successfully controlled

d

different devices in the cooling system and reduced peak power demand in operation.

Based on an easily understood programming language ladder programming, PLCs

te

can be programmed by users without advanced knowledge of programming [15].

Ac ce p

Moreover, since the programming of PLCs does not involve building dynamics and the prediction of building thermal behavior, the usage of PLCs for BITES is extremely flexible: they can easily suit the change of building operation and building-related systems; the only action needed is to input different instructions into PLCs. However, the fact that PLCs are based mainly on predetermined instructions and programming techniques, instead of the prediction of building thermal behavior, also makes it impossible for PLCs to achieve optimization of building operation.

2.2 Hard control 2.2.1 Model predictive control (MPC) MPC is defined as ―a very ample range of control methods which make an explicit use of a model of the process to obtain the control signal by minimizing an objective function‖ [16]. Particularly, when applied to building-related systems, it uses

Page 7 of 42

corresponding system dynamic models such as building energy simulations models to predict system thermal behavior. This prediction is then combined with optimization algorithms in order to determine the optimal control inputs. The MPC has three basic

ip t

elements:

(1) Explicit predictive models which can capture system dynamics. For building-

cr

related systems, the models could be system simulation models such as TRNSYS and Energy Plus, or other nonlinear system models established based on artificial

us

intelligence techniques such as artificial neural networks [17, 18],

(2) Objective function of which the MPC is to minimize or maximize. For building-

consumption and utility costs, and

an

related systems, typical objective functions include the total building energy

M

(3) Control law that minimizes the objective function.

MPC has been widely used to control active BITES [19, 20], passive BITES [21, 22],

d

and combined passive and active BITES [1, 23-25]. For example, Henze et al. [1] used both passive systems (precooling a building‘s massive structure) and active systems (ice

te

storage) in a three-story office building; and employed the MPC to harmonize them in

Ac ce p

order to minimize the building energy demand. TRNSYS was used to predict the building cooling load profiles based on weather forecast. The results show that significant utility cost savings and on-peak electrical demand reductions were achieved. Oldewurtel et al. [26] reported the main reasons of increasing the MPC applications in current buildings with and without TES is to reduce energy cost and to take advantage of time-of-use electricity rates.

The main advantage of MPC lies in its relative simplicity that can be understood by

engineers without advanced knowledge of control. Moreover, successful practical applications have already demonstrated its potential. The main barrier to the MPC application to BITES is the need for an accurate system dynamic model for prediction since the controller‘s performance is significantly affected by model accuracy. However, high accurate models are difficult to be developed considering building systems

Page 8 of 42

complexity [28], especially when a TES is integrated. Although many simulation tools (e.g., Energy Plus and TRNSYS) are available for MPC, they are not viewed as a control-friendly way due to their long run time, difficulties in communicating between them and optimization

ip t

algorithms, and difficulties of re-initialization of the developed model in terms of control feedback [27]. Efforts have been made recently in order to overcome the above barrier.

cr

For example, Wetter et al. [28, 29] developed a freely available Modelica library of building components for building energy and control systems in order to overcome the

us

limitations of traditional building simulation programs. Coffey et al. [30] developed a framework that simplifies the process of optimization with building energy simulation in

an

the MPC. However, further studies are still required and the focus could be the identification of the effect of model prediction performance on control performance

M

during practical system operation.

2.2.2 Optimal control

d

Optimal control techniques emerged in the 1950s as a mathematical method for

te

dealing with optimization problems through controlling engineering devices [31]. The optimal control is defined as ―to determine the control signals that will cause a process to

Ac ce p

satisfy the physical constraints and at the same time minimize (or maximize) some performance criterion‖ [32]. For BITES applications, the optimal control is a complex function of utility rates, load profiles, storage characteristics and weather parameters [6, 33, 34]. It is used to find a combination of different operating modes during specific periods to achieve minimal energy consumption and/or costs [35-37].

Braun [38] and Henze et al. [34] used the optimal control technique to control an ice

storage system, and used the storage charging mode, storage discharging mode and direct chiller mode as operating modes. Henze et al. [39] and Cheng et al. [40] investigated the optimal control of building thermal mass storage for time-of-use electric utility rates structures and identified the primary factors influencing the control performance. Hajiah and Krati [23, 24] developed the optimal control for both building thermal mass storage and ice storage to reduce the total operating costs including energy and demand charges

Page 9 of 42

while maintaining adequate occupant comfort conditions within commercial buildings. Optimal control is an efficient tool to improve the energy efficiency and reduce the operation cost of BITES. However, it suffers from complex computational burden

ip t

(especially for real-time applications) due to the model complexity and the operating

cr

condition variations.

2.2.3 Adaptive control

us

MPC essentially needs an accurate system model to make a reasonably accurate prediction. However, such an accurate model is not always available, particularly when system dynamics is complex. As an alternative, adaptive control techniques can be used

an

in order to overcome this shortcoming. de Silva [41] defined it as ―an adaptive control system is a feedback control system in which the values of some or all of the controller

M

parameters are modified (adapted) during the system operation (in real time) on the basis of some performance measure, when the response (output) requirements are not satisfied‖.

d

Clearly adaptive controllers with varying parameters are developed in order to accommodate changing dynamics of controlled systems. They can be classified into

te

different categories according to the methods of adjusting controller parameters and dealing with the dependency on accurate system models. For example, Wang et al. [42]

Ac ce p

developed rule-based adaptive controllers by extracting rules from typical system behavior, while Landau [43] developed model-referenced adaptive controllers (based on a reference model which outputs are used as the desired system response. Note that a reference model can be established only when controller designers are sufficiently familiar with system dynamics. The process of model-referenced adaptive control is illustrated in Fig. 2.

Page 10 of 42

Inputs

ip t an

-

us

Error signal

cr

Adaption

Reference model

Outputs

System operation

Controller

M

Fig. 2 Model-referenced adaptive control [43]

Fig. 2 shows that a reference model is used to provide desired outputs in response to reference inputs that are applied to both system operation and the reference model.

d

This model is selected to specify the required building performance, and is different from

te

the complex system models as developed for MPC. Its outputs are compared with realtime system operation outputs and the difference between them is the error signal. This

Ac ce p

error signal is then used to adjust controller parameters in order to reduce the error to zero. Clearly, the main advantage of model-referenced adaptive control is that an accurate system model is not necessary.

Similar to MPC, adaptive control can be combined with optimization algorithms in

order to achieve adaptive optimal control consisting generally of three basic elements as well [44]:

(1) Optimal control models, which mainly include objective functions and constraints, (2) Controller parameter identification, referring to the adjustment for time-varying adaptive controller parameters, and (3) Optimal algorithms, which is used to determine the optimum values for control variables.

Page 11 of 42

Wang [45] employed adaptive control and derivative control for a seawater-cooled chilling system online optimization. The two control techniques were utilized to determine seawater pressure set-points. Particularly, adaptive control is used to identify

ip t

the parameters used in derivative control that resets the pressure set-point. To evaluate the control performance, a chiller system of an existing office building was modeled and

cr

simulated. The results show that using these control techniques saved a significant

us

amount of energy.

The usage of adaptive control techniques in BITES is still rare. One possible reason

an

is the excessive control effort and high computational demand in calculation, particularly in large and complex building-related systems. For example, the usage of modelreferenced adaptive control places considerable effort during control process considering

M

referenced models are independent of real system models [41]. This makes it almost impossible to improve the prediction accuracies by using adaptive control [13].

d

Regarding rule-based adaptive control, its learning process commonly takes an unacceptably long time, which prevents its practical applications [46, 47]. Therefore,

te

further research is needed which should focus on the identification and dealing with the

Ac ce p

challenge of applying adaptive control to practical systems.

2.3 Soft control

In the past several decades, soft control has been proposed to address the challenge

of establishing accurate mathematical models to represent and predict system behavior analytically. Soft control can be viewed as a group of control techniques that combine conventional control and artificial intelligence (AI) methods such as neural networks and fuzzy logic.

2.3.1 Neural network (NN) control NN has been widely used in building-related systems with the ability of detecting, reproducing and predicting the complex nonlinear relationships between system input and

Page 12 of 42

output variables [48]. This enables it to learn complex system thermal behavior, which then is used in system control processes. For example, Liang and Du [49] designed a NN controller based on back-propagation algorithms in order to maintain the indoor comfort

ip t

level in terms of Predicted Mean Vote (PMV). Fig. 3 shows its basic structure.

NN controller

11

t

Building and HVAC system

us

1 Fixed input (bias)

Outputs

W12 Weight

For building control

Human thermal comfort indices calculation

M

Actual PMV Calculated based on measure d data

Measured data

an

Error derivative

Bi W as 13 te rm

-

W igh e W

Inputs + Reference PMV

cr

Error

Fig. 3 Basic structure of the designed NN control [49]

d

This two-layer NN controller has two inputs (i.e. error between Reference PMV and

te

actual PMV, error derivative) and one output (i.e. control signal to the building and

Ac ce p

HVAC system). The control procedure can be briefly described as:

Step 1: Give the weights random values to obtain initial inputs and outputs, Step 2: Update the weights based on the gradient descent algorithm, Step 3: Calculate the outputs based on the new weights and then send the control signals to the building and HVAC system.

Massie [50] utilized the NN control to harmonize a chiller with an ice storage tank

in order to minimize operating costs. The controller consists of a training network, used to learn the relationship between various cooling system parameters such as the chiller energy consumption, and a predictor network, used to determine control variable values such as chiller set-point temperatures and ice tank valve positions. The training network updates the NN weights and transfers them to the predictor network that determines

Page 13 of 42

control actions. Massie [50] concluded that a great advantage of the NN control over other control techniques, such as MPC, is the ability to learn patterns from system operational data, and thus take system variations like building retrofits into consideration. Moreover, system and component models provided by manufacturers can be significantly

ip t

different from field equipment due to their high non-linearity and their different installation locations. This requires considerable effort to fine-tune the control system.

cr

The limitation can be overcome by using the NN control that is able to self-calibrate those models. In addition, accurate system models are not required, which enables the

us

NN control to be implemented more easily in practice. However, normally a large

an

number of system operational data are needed to train the NN.

2.3.2 Fuzzy logic control

Fuzzy logic control is designed based on fuzzy logic that analyzes analog input

M

values instead of discrete input values. Fig. 4 shows the block diagram of a fuzzy

d

controller embedded in a closed-loop control system [51].

Rule-base

Defuzzification

te

Inference mechanism

Ac ce p

Reference inputs r(t)

Fuzzification

Fuzzy controller

Outputs y(t)

Inputs u(t)

Process

Fig. 4 A fuzzy controller embedded in a closed-loop control system [51]

The fuzzy controller consists of four components: (1) The fuzzification maps the input into a fuzzy value. For example, for the ‗indoor air temperature‘ input, 30 °C can be mapped into ‗a little high for occupants‘ while 40 °C can be mapped into ‗very high for occupants‘;

Page 14 of 42

(2) The rule-base includes a set of ―If-Then‖ rules predetermined by experts with relevant knowledge on system dynamics. For instance, one rule a building engineer may use is ‗If indoor air temperature is a little high then slightly turn down room thermostats‘; (3) The inference mechanism determines the rules to be used based on the input and then

ip t

defines the result of the rules; and

(4) The defuzzification converts the results into a crisp value. For example, ‗slightly turn

cr

down room thermostats‘ can be defuzzified as ‗turn down by one degree‘.

us

LeBreux et al. [52] used this technique to shift heating demand to off-peak periods. As PID control cannot deal with TES system‘s disturbances such as solar radiation

an

variations and thus often resulted in room overheating, Lebreux et al. [12] proposed a control model comprising a fuzzy logic controller and a feed-forward controller to overcome these drawbacks. The fuzzy logic controller was utilized to determine the

M

amount of electric energy to be stored in the TES system for the following day at each midnight. The results show that, compared to PID control, the proposed control can

d

maintain a comfortable thermal environment at all times without overheating while most

te

energy was consumed during off-peak hours.

Fuzzy logic control has an ability to deal with nonlinearities and uncertainties while

Ac ce p

an accurate system model is not necessarily required. Specifically, experts‘ experience can be used in the control design. Moreover, by adding new fuzzy rules or other new features, fuzzy logic controllers can be easily upgraded in order to make better-informed decisions. However, prior knowledge of system operation is required when defining fuzzy rules, which is a disadvantage.

2.4 Other control techniques 2.4.1 Reinforcement learning control Reinforcement learning is an area of machine learning concerned with how an agent should take actions in an environment in order to maximize cumulative reward, normally a long-term goal over a finite or infinite sequence of decisions [53]. The basic reinforcement learning model includes

Page 15 of 42

(1) A set of environment states S, (2) A set of actions a, (3) Rules of transitioning between states,

ip t

(4) Rules that determine the scalar immediate reward of a transition, and

cr

(5) Rules that describe what the agent observes.

As described in [46, 47], at any moment, t, the agent first senses the current

us

condition or state of the environment, denoted as St, and then selects an action, at. The selected action, at, leads to a new state, St+1, and after the state transition, the agent

an

receives a reward, rt. In a reinforcement learning problem, an agent interacts with the external environment to achieve a predefined long-term goal. The study of Henze and Schoenman [54] shows that the reinforcement learning control can effectively learn how

M

to control TES and achieved good performance. However, it was sensitive to the

d

selection of state variables, level of discretization, and learning rate.

Liu and Henze [46, 47] investigated the reinforcement learning for optimal control

te

of active and passive building thermal storage inventory. Through experimental study,

Ac ce p

they showed that this control can provide reliable control unitizing both active and passive thermal storage inventories and achieve a better performance than a variety of other control techniques, such as an optimal control of passive thermal storage with storage priority and a model-based optimization of active and passive thermal storage. The disadvantages of the proposed control are that the performance is sensitive to the quality of the simulator or the training model and the learning parameters, and it may also suffer from the curse of dimensionality of the state and action space.

2.4.2 Multi-agent control Multi-agent control, which has been introduced in Computer Science for decades, is under development in building systems and could be an outlook for the BITES. In such a system an intelligent supervisor to guide and coordinate the operation of the partial subsystems (i.e. local intelligent field controllers–agents) is necessary. Considering its ability

Page 16 of 42

to deal with the control and optimization of complex systems, multi-agent control can improve the control flexibility and performance when occupants, building owners and environmental agents are used to represent the building and its subsystems. This is because these agents‘ internal models not only allow them to compute the energy demand

ip t

and the associated costs, but also take account of the features of the thermal systems. Particularly, it is impossible to achieve optimal operation of local controllers without using

cr

individuals who are expert operators. Therefore, the architecture of a multi-agent control

us

system becomes necessary that will incorporate the knowledge of such expert operators.

Multi-agent control has been utilized in renewable-energy power-generation units

an

with electrical energy storage systems to improve operation efficiency and stability [55]. However, its application to BITES is still rare. Control strategies for the integration of TES into building systems should be developed within the framework of overall building

M

control strategies, both existing and planned. One issue deserved extra attention is to analyze and process complex and uncertain interactions between various energy-efficient

2.5 Hybrid control

te

d

components and TES. Multi-agent control could be a possible solution to this issue.

Ac ce p

Different control techniques have their own strengths and weaknesses. It is highly

desirable to combine them and develop hybrid controllers so that they can complement each other. For example, Liu and Henze [46, 47] developed a hybrid controller based on the MPC and reinforcement learning control. The hybrid control was divided into two phases:

Phase 1: a preliminary learning phase. A model was developed in order to ‗roughly‘ train the controller, and it is not necessary to match the system behavior perfectly.

Phase 2: a refined learning phase. After learning the trained experience in Phase 1, the controller was applied to an actual system. It continued to learn and further improves its performance through the interaction with the actual system.

Page 17 of 42

The controller‘s learning process with a ‗rough‘ model in Phase 1 helps to avoid the unacceptably long learning time needed in pure reinforcement learning. Also, the refined learning process in Phase 2 helps to improve the controller‘s performance and find the optimum system operation conditions while no accurate system simulation model is

ip t

necessary.

cr

In order to analyze the controller‘s performance, they designed a hybrid controller and implemented it to a commercial building with both passive and active BITES (full-

us

scale laboratory facilities). The results demonstrate its feasibility. Compared with a base case in which no TES is used, the hybrid control technique achieved 8.3% cost savings. It

an

is worth to mention that the cost savings achieved by using the hybrid controller are still much lower than that achieved by using the individual MPC. However, the utilization of the hybrid control technique is easier than that of the MPC due mainly to the fact that an

M

accurate system model is unnecessary. This is also very important since control system

d

manufacturers are able to produce the controller without much effort in practice.

te

3 Control strategies for BITES

Various strategies have been developed for controlling the operation of different

Ac ce p

active BITES, passive BITES, and combined active and passive BITES. This section presents and discusses the application of these control strategies.

3.1 Controlling Strategies adopted in different active BITES 3.1.1 Ice storage for building cooling Ice storage is a widely-used active BITES utilized for the provision of building

cooling. With consideration of time-of-use electricity rates, ice is produced during offpeak periods (commonly nighttime) and melted to discharge during on-peak periods (commonly daytime). In general, ice storage and chillers are combined to meet the requirements of building cooling, and the key issue of control strategies is the control of sequence and share of cooling discharge between chillers and ice. Existing strategies for ice storage can be classified into the following four categories [2, 6, 38, 56, 57]:

Page 18 of 42

(1) Chiller-priority: chillers have priority over ice storage. Ice storage is only used when total chiller capacity is insufficient for building cooling demand, (2) Constant-proportion: no priority is given to chillers or ice storage. Both chillers

all the time (e.g., ice storage: 40%; chillers: 60%),

ip t

and ice storage provide a certain fixed fraction of cooling required by buildings

cr

(3) Storage-priority: ice storage has priority over chillers. During off-peak periods, chiller-priority control is employed. During on-peak periods, chillers provide a

us

fixed fraction of cooling with the goal of melting all the ice produced during last off-peak period, and

an

(4) Rule-based: a strategy based on a series of if-then rules. The objective is to minimize or maximize the use of ice storage in terms of the fact that whether it increases or decreases daily energy costs. Specifically, if it increases daily energy

M

costs, then minimize it; vice versa.

d

Comparative studies have been conducted to evaluate the performance of these four control strategies. For example, Henze et al. [56] developed a simulation environment in

te

order to determine the optimal control strategy to minimize operating cost for ice storage

Ac ce p

system. The developed simulation tool was used to compare the performance of the first three strategies. The comparison was made in terms of the total operating cost that combines demand and energy charges. The results show that the storage-priority control gave the best performance. Braun [38] obtained the same results from a similar survey while his research excluded constant-proportion control. Drees and Braun [57] developed and evaluated a rule-based optimal control strategy for ice storage systems. The proposed strategy combined the elements of chiller-priority control and storage-priority control in a way that results in near-optimal performance under all conditions. In terms of the monthly electrical costs, they compared the proposed strategy with the chiller-priority strategy and storage-priority strategy based on the daily and monthly simulation of cooling systems. The results show that the proposed strategy delivered the best performance.

Page 19 of 42

In order to provide an overview of the strengths and weaknesses of these four control strategies, a comparison between them is made based on the reviewed literature,

cr

ip t

as shown in Table 1.

Table 1 Comparison between the four control strategies Constant-

Storage-

Rule-

priority

proportion

priority

based

Ease of implementation

Easy

Easy

Difficult

Most difficult

Load-shifting performance

Low

Medium

High

Highest

Need weather/load forecast

No

No

Yes

Yes

Operate cost saving

Low

Medium

High

Highest

M

an

us

Chiller-

d

3.1.2 HTWs for space heating and domestic hot water

te

HWTs are widely used as thermal energy storage systems to store heat for space heating and for domestic hot water (DHW) [58, 59]. This demand can be provided either

Ac ce p

by using two distinct systems or by using one combined system in cold areas. Combined systems have been increasingly used since they require less boilers or heat pumps. Moreover, PCMs are often integrated with HWTs in order to improve the heat storage capacity. In this article control strategies employed in combined systems with and without PCMs are reviewed separately.

3.1.2.1 HWTs without PCM Different types of combined systems have been developed and utilized. FernándezSeara et al. [60] grouped these systems into three broad categories:

(1) Space heating systems with indirect DHW heating systems, (2) DHW heaters with indirect space heating systems, and (3) Heat pump systems which provide space heating water and DHW.

Page 20 of 42

For these systems, usually hot water temperature both inside and outside HWTs needs to be controlled. A basic strategy for controlling hot water temperature inside HWTs is to turn off heating elements when the temperature reaches a predefined value

ip t

such as 65 °C i.e. on-off control. Considering that thermal stratification (i.e. hot water at the top and cold water in the bottom due to the gravity and buoyant effect) exists, such

cr

temperature is measured in a predetermined place.

us

Strategies for controlling hot water temperature outside HWTs are implemented with the ultimate goal of supplying hot water at a set-point temperature (e.g., 50 °C for

an

domestic water or 40 °C for floor heating water) through adjusting other parameters which mainly are water flow rates in practice. Since the amount of hot water provided is normally determined by load profiles, the major difference between adopted control

M

strategies is the methods of adjusting water flow rates.

d

Fernández-Seara et al. [60] classified space heating systems with indirect DHW

te

heating systems into three sub-categories:

Ac ce p

(1) DHW on demand that is produced based on hot water from tank through a heat exchanger,

(2) DHW produced based on hot water from DHW storage tank, and (3) Tank-in-tank systems. A DHW storage tank is placed inside a larger space heating tank.

They conducted experiments on the first sub-category of system and tested four

typical strategies for controlling supply DHW temperature, as illustrated in Fig. 5. In this diagram, 1, 2, 3 and 4 represents supply water from the space heating water tank, return water to the space heating water tank, supply DHW and return DHW, respectively. The heat exchange between space heating water and DHW occurs in the heat exchanger. The experiments were carried out on DHW load profiles of 5, 10, 15 and 20 l/min at 45 °C and the water in the space heating tank up to 80 °C.

Page 21 of 42

ip t cr us an M d te Ac ce p

Fig. 5 Domestic hot water production systems with four typical control strategies [60]

The dash line in the diagram indicates that the flow switch activates the pump when detecting DHW demand, thereby transferring heat through the heat exchanger. The four strategies for regulating supply DHW temperature are explained as follows:

A: Mixing supply DHW water with return DHW water through a 3-way valve, B: Adjusting the water flow rate of return water to the storage tank through a 2-

Page 22 of 42

way valve, C: Adjusting the flow rate of supply water from the water storage tank through a 2-way valve, and D: Adjusting both the flow rate and the temperature of supply water from the

ip t

water storage tank by mixing supply water with return water through a 3-way

cr

valve.

The results indicate that Strategy A and B provided DHW at the required

us

temperature stably throughout the range of DHW draw-off flow rates, whereas an unstable behavior of the system was found in Strategy C and D for draw-off flow rates of 5 l/min. Further analysis revealed that the control strategy also strongly affects the

an

formation and maintenance of thermal stratification in the space heating HWT. In summary, the control strategy exerts an important influence on both the quality of the

d

3.1.2.2 Inclusion of PCM in HWTs

M

DHW supply and the thermal performance of the space heating HWT.

te

HWTs store sensible heat in water for short term applications. However, the comparatively low heat storage capacity of sensible heat storage materials (i.e., water in

Ac ce p

this case) may not meet the requirements of load-shifting when peak period heating demand is high. In order to overcome this limitation, PCM modules are added into HWTs with the goal of taking advantage of their high latent heat value and isothermal operation [61, 62].

Many studies have been conducted in order to investigate the influence of adding

PCMs to HWTs on thermal behavior and system efficiency. These studies mainly focused on the effects of various influencing factors such as solar radiation intensity [63], PCM categories [61, 64, 65], PCM locations and thermal stratification [61, 62, 66], PCM quantity [61, 67, 68], and system configuration [69, 70]. A number of studies were also performed in order to develop numerical models for simulating the operation of HTWs integrated with PCM modules through TRNSYS [71, 72]. The primary focus of these studies has been the variation in average tank water temperature caused by the addition of

Page 23 of 42

PCM based on simple artificial hot-water load profiles. Control strategies for meeting the hot-water load requirements in practical applications, however, are seldom studied and tested. The focus of future research should be on the development and evaluation of control strategies for dealing with the complexity of actual hot-water load profiles and

ip t

taking advantage of the influence of PCMs on heat storage and thermal stratification

cr

enhancement.

3.1.3 Man-made heat/cool sources and building floors/roofs integrated with PCMs

us

Man-made heat/cool sources such as electric heaters have been added to building floors/roofs integrated with PCMs in order to increase and actively control heat storage.

an

The ultimate goal is to satisfy the occupants‘ requirements of thermal comfort and minimize the total building energy consumption. Achieving this goal depends highly on

M

adopted control strategies.

3.1.3.1 Under floor electric heating system In order to shift peak period heating demand to off-peak periods, under floor electric

d

heating systems incorporating PCM have become an effective and popular method [8, 73,

te

74]. In the heating season, this system can charge the PCM layer with the electrical heater during off-peak periods and discharge the heat to the floor tiles and then to indoor air

Ac ce p

during peak periods. The system performance is basically determined by strategies for controlling the electrical heater based on PCM characteristics and heating load profiles. Cho and Zaheer-uddin [75] reported that generally two categories of control strategies were adopted:

(1) Continuous heating strategy. The electric heater is controlled in terms of indoor environmental parameters. Normally feedback signals from indoor thermostats are used for the control. For example, based on the on-off control technique, the heater is turned off if indoor air temperature is higher than 20 °C while turned on if the temperature is lower than 20 °C, and (2) Intermittent heating strategy. The electric heater operates intermittently according to predetermined schedules (e.g., 3 times a day and 2 hours each time). No feedback control over indoor environmental parameters such as indoor air

Page 24 of 42

temperature or floor surface temperature is used.

The main advantage of continuous heating strategies lies in their ability to maintain indoor air temperature within an acceptable range. For example, Li et al. [74] developed

ip t

a form-stable PCM used in under floor electric heating systems. In order to evaluate its temperature-regulating and cost-reduction effects, simulation studies were performed

cr

based on different continuous heating strategies and PCM thicknesses. Three continuous heating strategies were defined and used in terms of indoor air temperature, floor surface

us

temperature, and both of them, respectively. Room temperature swings were considered an indicator of temperature-regulating effects. The results show that small temperature

an

swings were achieved by these heating strategies, which indicates a stable indoor air temperature.

M

The main advantage of intermittent heating strategies, compared with the continuous heating strategy, lies in their simplicity and high energy efficiency [74]. Currently most

d

intermittent heating strategies in practical applications are developed according to building designers‘/operators‘ previous relevant experience [76]. However, strategies

te

based on such experience may lead to unacceptably large indoor temperature swings

Ac ce p

considering thermal behavior of different buildings vary significantly, particularly when PCM layers are also integrated. One possible way to remove this barrier is to yield rational heating time based on real-time outdoor air temperature. For example, Cho and Zaheer-uddin [76] combined intermittent control and predictive control that based on the forecast of hourly outdoor air temperature, in order to determine the optimal number of daily heating hours. The results show that 10-12% energy was saved during the cold winter months by using the new control strategy.

In addition, to level heating energy demand, PCMs were also integrated into radiant floor systems with small tubes of hot water embedded in concrete floors. Similar to under floor electric heating systems, the control strategies for supplying hot water in tubes are crucial in meeting the requirements of peak shifting, energy cost and maintaining indoor thermal comfort. A typical strategy is to supply constant temperature hot water and

Page 25 of 42

control the water flow rate in terms of indoor air temperature. For example, Mazo et al. [77] developed a model to simulate radiant floor systems integrated with PCMs in a zone of a building, and validated it with experimental data. Based on the model, they analyzed the thermal behavior of a PCM radiant floor system fed by a heat pump with a constant

ip t

water supply temperature at 33 °C. The heat pump mainly operated at nighttime and thus took advantage of cheap electricity. On-off control with an indoor air temperature set-

cr

point of 21 °C was utilized for the heat pump operation. The results show that, with this strategy, the PCM enables to shift the peak period heating load to the off-peak periods

us

and an energy cost saving of nearly 18% is achieved compared to the system without

an

PCMs.

3.1.3.2 Building roofs incorporating PCMs PCMs have been integrated into building roofs mainly for cooling with the goal of

M

shifting energy peak as well as reducing the frequency of indoor air temperature swings. Cold water/air is used when PCMs are not able to completely solidify after nighttime cooling. Generally coolness of cold water/air is transferred and stored in PCMs before

te

periods.

d

peak demand periods; and then it is released to cool the room during peak demand

Ac ce p

Pasupathy et al. [78] conducted both experimental investigation and numerical

simulation to study the effects of having PCM panels on building roofs. The PCM charging process occurs during sunshine hours and it helps prevent heat from being transferred to indoor air. Then, during night hours, the PCMs solidify and release the heat to the ambient as well as the indoor air. Water pipes are installed inside the PCM panel so that cold water can be circulated through the PCM panel in the summer months. During the experiment a simple control strategy for the cold water circulation was implemented: the water was passed through the tubes for half an hour at 4 PM each afternoon. The results show that it can help maintain the roof bottom temperature at a constant level of 27 °C. However, a substantial quantity of water is required to extract the heat from the PCMs, which may give rise to difficulties for building designers as impractical. To remove this barrier, Pasupathy and Velraj [79] proposed a passive system with one more

Page 26 of 42

layer of PCMs with different phase change temperature above the bottom PCM panel, which is further introduced in Section 3.2.2.

3.1.4 PCM-based HVAC systems for building cooling

ip t

In order to meet building cooling requirements, PCMs have been incorporated into HVAC systems as cold storage mediums. According to [80-82], such system can be

cr

divided into three general categories:

us

(1) Free cooling systems,

(2) Air-conditioning systems integrated with PCMs, and

an

(3) Sorption (both absorption and adsorption) cooling systems integrated with PCMs.

M

3.1.4.1 Free cooling systems Free cooling is to integrate TES, mainly PCMs, into mechanical ventilation systems, thereby taking advantage of the coolness of the nights and then discharging it to rooms

d

during the daytime. Several reviews on free cooling of buildings were conducted in order to investigate where and how PCM was used as well as its effects on building energy

te

performance improvement [80, 83].

Ac ce p

In order to maintain indoor air temperature within a comfort range, common

strategies for controlling free cooling systems is to first switch between outdoor ventilation and indoor ventilation, and then adjust supply air flow rate. For example, Kang et al. [84] controlled the air flow rate in terms of three ventilation schemes:

(1) Outdoor ventilation scheme. Outdoor air is brought into the room through the PCM by the fan. When outdoor air temperature is lower than the PCM‘s melting temperature and the PCM has not been completely solidified, it will store ―coolness‖ to the PCM and also decrease room temperature. When outdoor air temperature is higher than the PCM‘s melting temperature, it will extract ―coolness‖ from PCM and then decrease room temperature, (2) Indoor ventilation scheme. When outdoor air temperature is lower than the PCM‘s

Page 27 of 42

melting temperature and the PCM has been completely solidified, no outdoor air is brought into the room and only the indoor air is circulated between the PCM and the room, and

ip t

(3) Stop fan scheme. The fan is stopped.

3.1.4.2 Air-conditioning (AC) systems integrated with PCMs PCMs have been integrated into AC systems with the goal of effectively meeting the

cr

requirement of energy redistribution. Bruno et al. [85] evaluated the thermal performance

us

of AC system integrated with PCMs through simulating annual building energy consumption. They assumed that building cooling load less than 2 kW would be tolerable and thus the cooling system was only turned on if cooling load was higher than 2 kW.

an

Based on this assumption, a control strategy was developed to meet building cooling load

M

demand, as follows:

(1) During off-peak load conditions between 22 PM and 7 AM, if cooling was required then the chiller would satisfy the requirement directly, otherwise the

d

chiller was used to charge the PCM system,

te

(2) During the daytime cooling was provided by the PCM system until it was fully melted or the temperature of returned chilled water was higher than 15 °C, then

Ac ce p

the chiller would be used to satisfy the requirement directly.

Parameshwaran and Kalaiselvam [86, 87] also proposed a system combining AC

and hybrid nano-composite particles embedded PCMs. They analyzed the system performance for summer and winter design conditions based on the following control strategy:

(1) During off-peak load conditions, a chiller operated at 80% of its nominal capacity to provide the building‘s cooling requirements and to charge the PCM system, (2) During on-peak load conditions, the chiller operated at 90% of its nominal

cooling capacity to meet the building‘s cooling requirements while the remaining requirements were met by stored coolness in the PCM system.

Page 28 of 42

3.1.4.3 Sorption cooling systems integrated with PCMs As an attractive alternative to conventional electrical chiller systems, sorption cooling systems, particularly LiBr/H2O absorption cooling systems, nowadays have been

ip t

widely used in buildings with the primary objective of reducing operation costs by avoiding peak electric demand charges. In order to improve its COP, in the past decade several studies have reported the experimental work to integrate PCMs into either

cr

condensers [88, 89] or generators [90]. These studies indicated that the combination of

us

PCMs and dry air coolers can further lower cooling water temperature, thereby achieving better system performance particularly in hot days. For example, Helm et al. [89] investigated the performance of four pilot installations of solar cooling and heating

an

systems with absorption chillers and PCMs. The nominal cooling capacity ranges from 7

Ac ce p

te

d

M

kW to 90 kW. The system structure in cooling mode is shown in Fig. 6.

Fig. 6 System structure of a solar cooling system with absorption chiller and latent heat storage in cooling mode [89] PCMs were integrated into the coolant loop and it was necessary to maintain cooling water temperature at approximately a constant temperature (i.e., 32 °C) before returning to the chiller. In order to achieve this goal, a valve was used after the dry cooler and corresponding control strategies were developed and implemented.

Page 29 of 42

(1) During the charging period, the valve is adjusted to control the water flow through the PCMs, thereby regulating the cooling water return temperature, (2) During the discharging period, with the goal of completely releasing the stored heat in PCMs, a parametric table of mass flow, fan speed and start time was pre-

ip t

determined through experiments and simulation. Thus different strategies could be implemented according to this parametric table based on outdoor air

cr

temperature.

us

3.2 Passive systems

PCMs have been integrated into different parts of building envelopes in order to

an

shift demand, and enhance indoor thermal comfort. These systems are defined as passive BITES since thermal energy are stored naturally without need of mechanical equipment

M

to deliver thermal energy. However, in order to maintain indoor air temperature within a comfort range, commonly HVAC systems as well as control strategies are still necessary in the cooling and heating seasons. An effective strategy can help to take full advantage

te

d

of PCM applications and maximize energy savings. 3.2.1 Building wallboards

Ac ce p

In the recent years there have been an increasing number of researches on the

integration of PCMs into building wallboards, which can be divided into three main groups:

(1) Develop new kinds of PCMs for energy-storing wallboards [91, 92], (2) Identify optimal design parameters of PCM wallboards such as PCM locations [93, 94], thicknesses of PCM boards [95], and optimal melting ranges of PCMs [96],

(3) Analyze PCM wallboards‘ thermal dynamics as well as their effects on thermal storage [96, 97], and (4) Develop design tools for optimally sizing the building envelope with PCM wallboards [98].

Many review articles on this topic have already been published during the last

Page 30 of 42

decade. Soares et al. [3] concluded that the research trend is towards more specialized studies such as the techniques to incorporate PCMs into building elements.

Although large amounts of experimental and numerical work have been presented,

ip t

generally the focus was placed on the use and effects of only PCM wallboards while HVAC systems and control strategies were seldom involved. In addition, most of them

cr

were carried out based on small cubicles or a single PCM wallboard room. This also restricts the use of HVAC systems. Recently several studies on PCM wallboards have

us

been conducted with HVAC systems utilized to regulate indoor air temperature. For example, Ascione et al. [99] simulated the hourly cooling energy consumption of a well-

an

insulated office building before and after the integration of PCM plaster into the inner side of exterior envelopes. They calculated the energy saving rate and the not-overheating time both with and without using air-conditioning systems. However, the objective is still

M

on maximizing energy savings through proper design and selection of PCMs, but not

te

3.2.2 Building Roofs

d

developing and evaluating different control strategies.

Few studies of building roofs integrated with PCMs have been conducted in the

Ac ce p

literature. Most of them focused mainly on the effect of PCMs on the reduction of heat flux through the roof into the house. HVAC systems and control strategies were seldom used to regulate indoor air temperature. For example, Alqallaf and Alawadhi [100] analyzed the thermal performance of a building concrete roof with vertical cylindrical holes filled with PCMs by using numerical techniques with an experimental validation. Similarly, Pasupathy and Velraj [79] investigated the effects of incorporating a doublelayer PCM panel in the roof. The top PCM has a melting temperature of 32 °C and the bottom PCM has a melting temperature of 27 °C. The results of both studies show that heat gain can be significantly reduced due to the fact that PCMs absorb the heat before it reaches the indoor space. However, it is difficult to maintain the indoor air temperature at a constant level within the comfort range, particularly in the hottest months.

Page 31 of 42

3.2.3 Building floors In general, building floors integrated with PCMs are used in passive solar buildings [101, 102]. These studies can be grouped into two categories:

ip t

(1) Researchers conducted experiments to investigate the effects of floors integrating with PCMs on space heating,

cr

(2) Researchers developed simulation models and then performed parametric studies to investigate the influence of various factors (e.g., PCM thicknesses and PCM

us

melting temperature) on room thermal performance.

an

Both categories of studies are carried out without applying HVAC systems and

3.2.4 Windows curtains and shutters

M

control strategies regulating indoor air temperature.

The concept of PCM-related window systems has been proposed and its effect on

d

reducing building heat gain/losses has been studied for decades. For example, PCMs

te

were used in window shutters in order to reduce cooling load in summer [103] or increase window heat storage capacity as well as indoor air temperature during nighttime [104,

Ac ce p

105]. Moreover, thermally efficient windows with double glass panels filled with PCMs were also investigated [106]. However, similar to above-mentioned passive BITES, the research on its applications with HVAC systems and control strategies for regulating indoor air temperature are seldom conducted.

3.3 Combined active and passive systems Both active and passive BITES have a marked effect on expanding the use of

renewables and shifting peak period building loads. In order to exploit their potential, it is highly desirable to combine them effectively. So far limited work has been performed on testing this combination, with the main focus on combining ice storage systems, which are charged at night and discharged during the day, with building thermal masses, which are pre-cooled during the night and early morning hours [1, 23-25, 46, 47, 107].

Page 32 of 42

In these studies optimal control strategies were employed for achieving maximum cost savings. Specifically, in the control process optimization algorithms were used with the goal of identifying the pre-determined control parameters which minimize certain

ip t

performance criterion such as total costs. The control parameters such as indoor air setpoints and TES charge/discharge rates are calculated and reset in real time based on

cr

ambient conditions and occupancy patterns. Such strategies are commonly implemented based on optimal control techniques, and can be further split into three sub-categories [46,

us

47]:

an

(1) Optimal control of the passive systems only. Building global zone air temperature set-point is used as the control variable that takes advantage of the building thermal storage through pre-cooling. Optimized zone air temperature set-points

M

are identified and updated by model-based optimization. The active systems are not used,

d

(2) Optimal control of the passive systems while the active systems are utilized and controlled by either storage-priority or chiller-priority strategy, and

te

(3) Optimal control of both the active and passive systems. Building global zone air

Ac ce p

temperature set-points and charging/discharging rates for the active systems are used as control variables. Both variables are optimized during operation by model-based optimization.

In order to evaluate the effects of the optimal control strategies on the building

energy performance, Hajiah and Krarti [23, 24] developed a simulation environment to estimate building energy use, peak demand, and thermal comfort when employing the strategies within commercial buildings. Based on the simulation environment, three optimal control tests were carried out with the objective function as energy costs, peak demand costs and total costs, respectively. Optimal values for both indoor temperature and charging/discharging rates for an ice storage system were identified during operation. For example, when minimizing the peak demand cost, the strategy adopted for controlling both the active and passive systems can be described as:

Page 33 of 42

(1) 5 PM (the end of occupied periods) - 8 AM (the beginning of occupied periods): the pre-cooling process started through maintaining the indoor air temperature at 15.5 °C,

ip t

(2) 8 AM - 11AM (off-peak occupied hours): two control modes were used in sequence:

8 AM - 9 AM: no conditioned air was supplied to the building so that the

cr



indoor air temperature was increased to 20 °C (i.e., the lower bound of the 

us

thermal comfort range),

9 AM - 11AM: the indoor air temperature was maintained at 20 °C,

an

(3) 11 AM - 5 PM (the on-peak period): the indoor temperature was gradually increased from 20 °C to 24.4 °C (i.e., the upper bound of the thermal comfort

M

range) through varying the airflow rate supplied to the building.

The results indicate that, compared with conventional control strategies, significant

d

cost savings could be achieved: 13.2% savings in demand charges and 10.8% savings in

te

total charges were obtained. However, the complexities of adopted optimization algorithms as well as heavy computation load caused by the search of optimal control

Ac ce p

values might impose severe limitations on the optimal strategies‘ wide applications in practice.

4 Conclusions

This paper presents a review of control techniques and strategies utilized for

integrating TES into different building systems. The concluding remarks and recommendations for future work in this area are as follows.  Owing to its relative simplicity and reliability, the PID controller is in practice the most adopted controller in BITES. It can either be used as a stand-alone controller or, more probably, be embedded in PLCs that can be programmed by users to store predefined instructions. However, it might be difficult for PLC-

Page 34 of 42

based PID controllers to achieve system performance optimization.  The requirements of system performance optimization have stimulated the research on the application of various hard control techniques, such as the MPC

ip t

and optimal control, to BITES. These techniques combine system dynamic models with optimization algorithms in order to determine the optimal control

cr

inputs or settings. Considerable difficulties in developing high quality models, particularly for complex BITES, results in the decrease of control performance.

us

As an alternative, adaptive control can be used to overcome this shortcoming. However, its practical applications to BITES are still rare and more tests need to

an

be conducted to assess its performance.

 Both NN and fuzzy logic techniques have an aptitude for dealing with

M

nonlinearities and uncertainties of BITES while an accurate system model is not necessarily required. Moreover, they can easily take system variations like building retrofits into consideration by learning patterns from system operational

d

data and by adding new fuzzy rules, respectively. However, a large number of

te

system operational data are normally needed for training the NN and prior

Ac ce p

knowledge of system operation is required when defining fuzzy rules.  Reinforcement learning control was also used to control BITES. The experiment shows that this model-free technique can achieve a better system performance than a variety of control techniques such as model-based optimal control. However, the learning time reduction will be the key issue to be addressed in future studies.

 Control strategies for the integration of TES into building systems should be developed within the framework of overall building control strategies, both existing and planned. One issue deserved extra attention is to analyze and process complex and uncertain interactions between various energy-efficient components and TES. Multi-agent control could be a possible solution to this issue.

Page 35 of 42

 It is highly desirable to combine different control techniques and develop hybrid controllers so that they can compensate each other. The feasibility of a hybrid

ip t

controller based on both the MPC and reinforcement learning control has already been demonstrated in previous studies on BITES. An accurate system model is

cr

unnecessary and learning time is reduced. But the system cost savings achieved by using the hybrid controller are much lower than that achieved by using the

us

MPC.

 Ice storage is usually combined with chillers to meet the building cooling

an

demands, while the operation priority needs to be determined in terms of different scenarios. Existing strategies for setting priorities include chiller-

M

priority, constant-proportion, storage-priority and rule-based. A trade-off between their strengths and weaknesses, such as the difficulty levels of implementation and the necessity of weather forecast, need to be taken into

te

d

consideration before the strategy adoption during the design stage.  A basic strategy for maintaining water temperature inside HWTs at a predefined

Ac ce p

set-point is to turn on or off heating devices by using on-off control. Strategies for controlling supply water temperature outside HWTs depend on heat exchange modes between space heating water and DHW. To improve the heat storage capacity PCMs are often integrated into HWTs while the associated control strategies are seldom studied and tested. The focus of future research should be on the development and evaluation of control strategies for dealing with the complexity of practical hot-water load profiles and taking advantage of the influence of PCMs on heat storage and thermal stratification enhancement.

 Man-made heat/cool sources have been added to building floors/roofs integrated with PCMs in order to increase and actively control heat storage. Both continuous and intermittent strategies were adopted in these systems. However, these strategies were developed based mainly on designers‘ experience and were

Page 36 of 42

not optimized. More research on optimal strategies development is needed to achieve load-shifting and energy efficiency improvement.  PCMs have been incorporated into free cooling systems, central AC systems and

ip t

sorption cooling systems as cold storage mediums. Control strategies adopted in these systems are to maintain indoor air temperature by adjusting supply air flow

cr

rate, to meet the cooling demand by determining the timing of chiller usage, and to maintain cooling water temperature at a lower temperature by adjusting the

us

water flowing through the PCM, respectively. However, the strategies were developed mainly based on the researchers‘ experience without adequate justification. A major problem to be addressed is how to determine the optimal

an

PCM charging/discharging timing and length, which are subject to various

M

influencing factors such as weather conditions and PCM properties.  As passive systems, PCMs have been integrated into different parts of building envelopes and no mechanical equipment is used. Most previous relevant studies

d

focused mainly on the thermal effect of PCM integration. In order to maintain

te

indoor air temperature within a comfort range and take full advantage of PCMs, it is suggested that HVAC systems as well as control strategies regulating indoor

Ac ce p

air temperature be employed in the cooling and heating seasons. Further research work in this regard is needed to first conduct parametric analyses for identifying key parameters that affecting overall system performance, and then develop effective strategies for controlling HVAC systems.

 So far limited work has been performed on testing combined active and passive BITES, with the main focus on combining ice storage systems with building thermal masses. Optimal control strategies were employed for achieving maximum cost savings. The complexities of adopted optimization algorithms as well as heavy computation load caused by the search of optimal control values might impose severe limitations on the optimal strategies‘ wide application in practice. Further research on addressing this issue is still needed. In addition, research on other combinations of different active and passive systems as well as

Page 37 of 42

associated control strategies is also needed.

Acknowledgement

ip t

This work was financially supported by the National Natural Science Foundation of China (No. 51408205), the Fundamental Research Funds for the Central Universities and

us

cr

Hunan Provincial Science and Technology Major Project of China (No. 2011FJ1007).

References

Ac ce p

te

d

M

an

[1] G.P. Henze, C. Felsmann, G. Knabe. Evaluation of optimal control for active and passive building thermal storage. International Journal of Thermal Sciences. 43(2)(2004) 173-183. [2] W.J. Cole, K.M. Powell, T.F. Edgar. Optimization and advanced control of thermal energy storage systems. Reviews in Chemical Engineering. 28(2-3)(2012) 81-99. [3] N. Soares, J.J. Costa, A.R. Gaspar, P. Santos. Review of passive PCM latent heat thermal energy storage systems towards buildings‘ energy efficiency. Energy and Buildings. 59(0)(2013) 82-103. [4] Y. Sun, S. Wang, F. Xiao, D. Gao. Peak load shifting control using different cold thermal energy storage facilities in commercial buildings: A review. Energy Conversion and Management. 71(0)(2013) 101-114. [5] S. Pintaldi, C. Perfumo, S. Sethuvenkatraman, S. White, G. Rosengarten. A review of thermal energy storage technologies and control approaches for solar cooling. Renewable and Sustainable Energy Reviews. 41(0)(2015) 975-995. [6] ASHRAE. ASHRAE Handbook-HVAC Application. American Society of HVAC Engineers Inc. 2011. [7] A. Afram, F. Janabi-Sharifi. Theory and applications of HVAC control systems – A review of model predictive control (MPC). Building and Environment. 72(0)(2014) 343-355. [8] K. Lin, Y. Zhang, X. Xu, H. Di, R. Yang, P. Qin. Experimental study of under-floor electric heating system with shape-stabilized PCM plates. Energy and Buildings. 37(3)(2005) 215-220. [9] K.M. Powell, T.F. Edgar. Control of a large scale solar thermal energy storage system. American Control Conference. San Francisco, CA, USA. 2011. [10] A. Mawire, M. McPherson. Experimental characterisation of a thermal energy storage system using temperature and power controlled charging. Renewable Energy. 33(4)(2008) 682-693. [11] A. Visioli. Practical PID Control. Springer. 2006. [12] M. LeBreux, M. Lacroix, G. Lachiver. Control of a hybrid solar/electric thermal energy storage system. International Journal of Thermal Sciences. 48(3)(2009) 645-654. [13] S.W. Wang, Z.J. Ma. Supervisory and Optimal Control of Building HVAC Systems: A Review. HVAC&R RESEARCH. 14(1)(2008) 3-32. [14] M.E.S. Farahani, N. Saeidi. Case Study of Design and Implementation of a Thermal Energy Storage System. Power and Energy Conference, PECon '06. Putrajaya, Malaysia. 2006. [15] W. Bolton. Programmable Logic Controllers. Fifth Edition. Oxford. Newnes. 2009. [16] E.F. Camacho, C. Bordons. Model predictive control. Springer. Berlin, New York. 1999. [17] C.E. Garcia, D.M. Prett, M. Morari. Moedel predictive control: theory and practice-a survey. Automation. 25(3)(1989) 335-348. [18] P. May-Ostendorp, G.P. Henze, C.D. Corbin, B. Rajagopalan, C. Felsmann. Model-predictive control of mixed-mode buildings with rule extraction. Building and Environment. 46(2)(2011) 428-437. [19] W.J. Cole, K.M. Powell, E.T. Hale, T.F. Edgar. Reduced-order residential home modeling for model predictive control. Energy and Buildings. 74(0)(2014) 69-77. [20] Y. Ma, F. Borrelli, B. Hencey, B. Coffey, S. Bengea, P. Haves. Model Predictive Control for the Operation of Building Cooling Systems. Transactions on control systems technology. 20(3)(2012) 796-803.

Page 38 of 42

Ac ce p

te

d

M

an

us

cr

ip t

[21] E.M. Greensfelder, G.P. Henze, C. Felsmann. An investigation of optimal control of passive building thermal storage with real time pricing. Journal of Building Performance Simulation. 4(2)(2011) 91-104. [22] J.R. Ma, J. Qin, T. Salsbury, P. Xu. Demand reduction in building energy systems based on economic model predictive control. Chemical engineering science. 67(1)(2012) 92-100. [23] A. Hajiah, M. Krarti. Optimal control of building storage systems using both ice storage and thermal mass – Part I: Simulation environment. Energy Conversion and Management. 64(0)(2012) 499-508. [24] A. Hajiah, M. Krarti. Optimal controls of building storage systems using both ice storage and thermal mass – Part II: Parametric analysis. Energy Conversion and Management. 64(0)(2012) 509-515. [25] J. Candanedo, A. Athienitis. Predictive control of radiant floor heating and solar-source heat pump operation in a solar house. HVAC&R RESEARCH. 17(3)(2011) 235-256. [26] F. Oldewurtel, A. Parisio, C.N. Jones, D. Gyalistras, M. Gwerder, V. Stauch, B. Lehmann, M. Morari. Use of model predictive control and weather forecasts for energy efficient building climate control. Energy and Buildings. 45(0)(2012) 15-27. [27] F. Sakellariou. Model Predictive Control for Thermally Activated Building Systems. Thesis. Eindhoven University of Technology. 2011. [28] M. Wetter. Modelica-based modeling and simulation to support research and development in building energy and controls systems. Journal of Building Performance Simulation. 2(2)(2009) 143-161 [29] M. Wetter, W.D. Zuo, T.S. Nouidui. Recent developments of the modelica "buildings" library for building energy and control systems. the 8th International Modelica Conference. Dresden, Germany. 2011. [30] B. Coffey, F. Haghighat, E. Morofsky, E. Kutrowski. A software framework for model predictive control with GenOpt. Energy and Buildings. 42(7)(2010) 1084-1092. [31] R. Vinter. Optimal Control. Springer. Birkhauser. Boston. 2010. [32] D.E. Kirk. Optimal control theory: an introduction. Prentice Hall, Inc. New York. 1971. [33] K. M., B.M. J., G.P. Henze. Evaluation of Optimal Control for Ice Storage Systems (RP-809). ASHRAE research report. Atlanta, GA. R. a. A.-C. E. American Society of Heating, Inc. 1996. [34] H. G.P., D. R.H., K. M. Development of a predictive optimal controller for thermal energy storage systems. HVAC&R RESEARCH. 3(3)(1997) 233-264. [35] J.E. Braun. Impact of control on operating costs for cool storage systems with dynamic electric rates. ASHRAE Transactions. 113(2)(2007) 343-354. [36] Z. Z.Q., T. W.D., C. Q., C. X., D. S. Methodology for determining the optimal operation strategies for a chilled-water-storage system-Part I: theoretical model. HVAC&R RESEARCH. 17(2011) 737-751. [37] Z. Z.Q., T. W.D., C. Q., C. X., D. S. Methodology for determining the optimal operating strategies for a chilled-water-storage system-Part II: Project application. HVAC&R RESEARCH. 17(2011) 751-770. [38] J.E. Braun. A comparison of chiller-priority, storage-priority, and optimal control of an ice system. ASHRAE Transactions. 98 (1)(1992) 893-902. [39] H. G.P., B. M.J., F. C., F. A., C. H. Evaluation of building thermal mass savings. ASHRAE Research Project RP-1313. 2007. [40] C. H., B. M.J., H. G.P., F. A., F. C. Evaluation of the primary factors impacting the optimal control of passive thermal storage (RP1313). . ASHRAE Transactions. 114(2)(2008) 57-64. [41] C.W.d. Silva. Intelligent control: fuzzy logic applications. CRC Press. New York. 1995. [42] H.Wang, G.P. Liu, C.J. Harris, M.T. Brown. Advanced Adaptive Control. Pergamon Press. 1995. [43] I.D. Landau. Adaptive control: The model reference approach. Marcel Dekker. New York. 1979. [44] Y. Yan, J. Zhou, Y. Lin, W. Yang, P. Wang, G. Zhang. Adaptive optimal control model for building cooling and heating sources. Energy and Buildings. 40(8)(2008) 1394-1401. [45] S. Wang. Dynamic simulation of a building central chilling system and evaluation of EMCS on-line control strategies. Building and Environment. 33(1)(1998) 1-20. [46] S. Liu, G.P. Henze. Experimental analysis of simulated reinforcement learning control for active and passive building thermal storage inventory: Part 1. Theoretical foundation. Energy and Buildings. 38(2)(2006) 142-147. [47] S. Liu, G.P. Henze. Experimental analysis of simulated reinforcement learning control for active and passive building thermal storage inventory: Part 2: Results and analysis. Energy and Buildings. 38(2)(2006) 148-161. [48] Z. Yu, F. Haghighat, B.C.M. Fung, H. Yoshino. A decision tree method for building energy demand modeling. Energy and Buildings. 42(10)(2010) 1637-1646. [49] J.Liang, R.X. Du. Thermal comfort control based on neural network for HVAC application. 2005 IEEE Conference on control applications. Toronto, Canada. 2005.

Page 39 of 42

Ac ce p

te

d

M

an

us

cr

ip t

[50] D.D. Massie. Optimization of a building's cooling plant for operating cost and energy use. International Journal of Thermal Sciences. 41(12)(2002) 1121-1129. [51] K.M. Passino, S. Yurkovich. Fuzzy Control. Addison Wesley Longman. Menlo Park, CA. 1998. [52] M. LeBreux, M. Lacroix, G. Lachiver. Fuzzy and feedforward control of an hybrid thermal energy storage system. Energy and Buildings. 38(10)(2006) 1149-1155. [53] S. Sutton, B. Richard, G. Andrew. Reinforcement Learning: an introduction. MIT Press. USA. 1998. [54] G.P. Henze, J. Schoenmann. Evaluation of reinforcement learning control for thermal energy storage systems. HVAC&R RESEARCH. 9(2003) 259-275. [55] K. Wu, H. Zhou. A multi-agent-based energy-coordination control system for grid-connected largescale wind–photovoltaic energy storage power-generation units. Solar Energy. 107(0)(2014) 245-259. [56] G.P. Henze, M. Krarti, M.J. Brandemuehl. A simulation environment for the analysis of ice storage controls. HVAC&R RESEARCH. 3(2)(1997) 128-148. [57] K.H. Drees, J.E. Braun. Development and evaluation of a rule-based control strategy for ice storage systems. HVAC&R RESEARCH. 2(4)(1996) 312-336. [58] S. Shao, W. Shi, X. Li, J. Ma. A new inverter heat pump operated all year round with domestic hot water. Energy Conversion and Management. 45(13–14)(2004) 2255-2268. [59] Y. Hamada, J. Fukai. Latent heat thermal energy storage tanks for space heating of buildings: Comparison between calculations and experiments. Energy Conversion and Management. 46(20)(2005) 3221-3235. [60] J. Fernández-Seara, F.J. Uhía, Á.Á. Pardiñas, S. Bastos. Experimental analysis of an on demand external domestic hot water production system using four control strategies. Applied Energy. 103(0)(2013) 85-96. [61] D.N. Nkwetta, P.-E. Vouillamoz, F. Haghighat, M. El-Mankibi, A. Moreau, A. Daoud. Impact of phase change materials types and positioning on hot water tank thermal performance: Using measured water demand profile. Applied Thermal Engineering. 67(1–2)(2014) 460-468. [62] D.N. Nkwetta, P.-E. Vouillamoz, F. Haghighat, M. El Mankibi, A. Moreau, K. Desai. Phase change materials in hot water tank for shifting peak power demand. Solar Energy. 107(0)(2014) 628-635. [63] M.A. Fazilati, A.A. Alemrajabi. Phase change material for enhancing solar water heater, an experimental approach. Energy Conversion and Management. 71(0)(2013) 138-145. [64] M. Mazman, L.F. Cabeza, H. Mehling, M. Nogues, H. Evliya, H.Ö. Paksoy. Utilization of phase change materials in solar domestic hot water systems. Renewable Energy. 34(6)(2009) 1639-1643. [65] S. Canbazoğlu, A. Şahinaslan, A. Ekmekyapar, Ý.G. Aksoy, F. Akarsu. Enhancement of solar thermal energy storage performance using sodium thiosulfate pentahydrate of a conventional solar water-heating system. Energy and Buildings. 37(3)(2005) 235-242. [66] D. Haillot, E. Franquet, S. Gibout, J.-P. Bédécarrats. Optimization of solar DHW system including PCM media. Applied Energy. 109(0)(2013) 470-475. [67] M. Nabavitabatabayi, F. Haghighat, A. Moreau, P. Sra. Numerical analysis of a thermally enhanced domestic hot water tank. Applied Energy. 129(0)(2014) 253-260. [68] L.F. Cabeza, M. Ibáñez, C. Solé, J. Roca, M. Nogués. Experimentation with a water tank including a PCM module. Solar Energy Materials and Solar Cells. 90(9)(2006) 1273-1282. [69] N.H.S. Tay, M. Belusko, F. Bruno. Experimental investigation of tubes in a phase change thermal energy storage system. Applied Energy. 90(1)(2012) 288-297. [70] A.J.N. Khalifa, K.H. Suffer, M.S. Mahmoud. A storage domestic solar hot water system with a back layer of phase change material. Experimental Thermal and Fluid Science. 44(0)(2013) 174-181. [71] M. Ibáñez, L.F. Cabeza, C. Solé, J. Roca, M. Nogués. Modelization of a water tank including a PCM module. Applied Thermal Engineering. 26(11–12)(2006) 1328-1333. [72] J. Tamasauskas, M. Poirier, R. Zmeureanu, R. Sunyé. Modeling and optimization of a solar assisted heat pump using ice slurry as a latent storage material. Solar Energy. 86(11)(2012) 3316-3325. [73] K. Lin, Y. Zhang, X. Xu, H. Di, R. Yang, P. Qin. Modeling and simulation of under-floor electric heating system with shape-stabilized PCM plates. Building and Environment. 39(12)(2004) 1427-1434. [74] J. Li, P. Xue, H. He, W. Ding, J. Han. Preparation and application effects of a novel form-stable phase change material as the thermal storage layer of an electric floor heating system. Energy and Buildings. 41(8)(2009) 871-880. [75] S.H. Cho, M. Zaheer-uddin. An experimental study of multiple parameter switching control for radiant floor heating systems. Energy. 24(5)(1999) 433-444.

Page 40 of 42

Ac ce p

te

d

M

an

us

cr

ip t

[76] S.H. Cho, M. Zaheer-uddin. Predictive control of intermittently operated radiant floor heating systems. Energy Conversion and Management. 44(8)(2003) 1333-1342. [77] J. Mazo, M. Delgado, J.M. Marin, B. Zalba. Modeling a radiant floor system with Phase Change Material (PCM) integrated into a building simulation tool: Analysis of a case study of a floor heating system coupled to a heat pump. Energy and Buildings. 47(0)(2012) 458-466. [78] A. Pasupathy, L. Athanasius, R. Velraj, R.V. Seeniraj. Experimental investigation and numerical simulation analysis on the thermal performance of a building roof incorporating phase change material (PCM) for thermal management. Applied Thermal Engineering. 28(5–6)(2008) 556-565. [79] A. Pasupathy, R. Velraj. Effect of double layer phase change material in building roof for year round thermal management. Energy and Buildings. 40(3)(2008) 193-203. [80] E. Osterman, V.V. Tyagi, V. Butala, N.A. Rahim, U. Stritih. Review of PCM based cooling technologies for buildings. Energy and Buildings. 49(0)(2012) 37-49. [81] A. El-Sawi, F. Haghighat, H. Akbari. Centralized latent heat thermal energy storage system: Model development and validation. Energy and Buildings. 65(0)(2013) 260-271. [82] A. El-Sawi, F. Haghighat, H. Akbari. Assessing long-term performance of centralized thermal energy storage system. Applied Thermal Engineering. 62(2)(2014) 313-321. [83] V.A.A. Raj, R. Velraj. Review on free cooling of buildings using phase change materials. Renewable and Sustainable Energy Reviews. 14(9)(2010) 2819-2829. [84] K. Yanbing, J. Yi, Z. Yinping. Modeling and experimental study on an innovative passive cooling system—NVP system. Energy and Buildings. 35(4)(2003) 417-425. [85] F. Bruno, N.H.S. Tay, M. Belusko. Minimising energy usage for domestic cooling with off-peak PCM storage. Energy and Buildings. 76(0)(2014) 347-353. [86] R. Parameshwaran, S. Kalaiselvam. Energy efficient hybrid nanocomposite-based cool thermal storage air conditioning system for sustainable buildings. Energy. 59(0)(2013) 194-214. [87] R. Parameshwaran, S. Kalaiselvam. Energy conservative air conditioning system using silver nanobased PCM thermal storage for modern buildings. Energy and Buildings. 69(0)(2014) 202-212. [88] M. Helm, C. Keil, S. Hiebler, H. Mehling, C. Schweigler. Solar heating and cooling system with absorption chiller and low temperature latent heat storage: Energetic performance and operational experience. International Journal of Refrigeration. 32(4)(2009) 596-606. [89] M. Helm, K. Hagel, W. Pfeffer, S. Hiebler, C. Schweigler. Solar Heating and Cooling System with Absorption Chiller and Latent Heat Storage – A Research Project Summary. Energy Procedia. 48(0)(2014) 837-849. [90] F. Agyenim, M. Rhodes, I. Knight. The use of phase change material (PCM) to improve the coefficient of performance of a chiller for meeting domestic cooling in Wales. The 2nd PALENC conference and 28th AIVC conference on building low energy cooling and advanced ventilation technologies in the 21st Century. Crete island, Greece. 2007. [91] C. Chen, H. Guo, Y. Liu, H. Yue, C. Wang. A new kind of phase change material (PCM) for energystoring wallboard. Energy and Buildings. 40(5)(2008) 882-890. [92] L. Shilei, Z. Neng, F. Guohui. Eutectic mixtures of capric acid and lauric acid applied in building wallboards for heat energy storage. Energy and Buildings. 38(6)(2006) 708-711. [93] X. Jin, M.A. Medina, X. Zhang. On the importance of the location of PCMs in building walls for enhanced thermal performance. Applied Energy. 106(0)(2013) 72-78. [94] S.D. Zwanzig, Y. Lian, E.G. Brehob. Numerical simulation of phase change material composite wallboard in a multi-layered building envelope. Energy Conversion and Management. 69(0)(2013) 27-40. [95] J. Koo, H. So, S.W. Hong, H. Hong. Effects of wallboard design parameters on the thermal storage in buildings. Energy and Buildings. 43(8)(2011) 1947-1951. [96] G. Evola, L. Marletta, F. Sicurella. A methodology for investigating the effectiveness of PCM wallboards for summer thermal comfort in buildings. Building and Environment. 59(0)(2013) 517-527. [97] G. Zhou, Y. Yang, X. Wang, J. Cheng. Thermal characteristics of shape-stabilized phase change material wallboard with periodical outside temperature waves. Applied Energy. 87(8)(2010) 2666-2672. [98] A. Bastani, F. Haghighat, J. Kozinski. Designing building envelope with PCM wallboards: Design tool development. Renewable and Sustainable Energy Reviews. 31(0)(2014) 554-562. [99] F. Ascione, N. Bianco, R.F. De Masi, F. de‘ Rossi, G.P. Vanoli. Energy refurbishment of existing buildings through the use of phase change materials: Energy savings and indoor comfort in the cooling season. Applied Energy. 113(0)(2014) 990-1007.

Page 41 of 42

Ac ce p

te

d

M

an

us

cr

ip t

[100] H.J. Alqallaf, E.M. Alawadhi. Concrete roof with cylindrical holes containing PCM to reduce the heat gain. Energy and Buildings. 61(0)(2013) 73-80. [101] X. Xu, Y. Zhang, K. Lin, H. Di, R. Yang. Modeling and simulation on the thermal performance of shape-stabilized phase change material floor used in passive solar buildings. Energy and Buildings. 37(10)(2005) 1084-1091. [102] A.G. Entrop, H.J.H. Brouwers, A.H.M.E. Reinders. Experimental research on the use of microencapsulated Phase Change Materials to store solar energy in concrete floors and to save energy in Dutch houses. Solar Energy. 85(5)(2011) 1007-1020. [103] E.M. Alawadhi. Using phase change materials in window shutter to reduce the solar heat gain. Energy and Buildings. 47(0)(2012) 421-429. [104] D. Buddhi, H.S. Mishra, A. Sharma. Thermal performance studies of a test cell having a PCM window in south direction. Annex 17. IEA. Energy conservation through energy storage. 2003. [105] T. Silva, R. Vicente, F. Rodrigues, A. Samagaio, C. Cardoso. Development of a window shutter with phase change materials: Full scale outdoor experimental approach. Energy and Buildings. 88(0)(2015) 110121. [106] K.A.R. Ismail, J.R. Henr quez. Thermally effective windows with moving phase change material curtains. Applied Thermal Engineering. 21(18)(2001) 1909-1923. [107] M. Kintner-Meyer, A.F. Emery. Optimal control of an HVAC system using cold storage and building thermal capacitance. Energy and Buildings. 23(1)(1995) 19-31.

Page 42 of 42