Energy Conversion and Management 64 (2012) 509–515

Contents lists available at SciVerse ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Optimal controls of building storage systems using both ice storage

and thermal mass – Part II: Parametric analysis
Ali Hajiah a, Moncef Krarti b,⇑
a
Building and Energy Technologies Department, Environmental & Urban Development Division, Kuwait Institute for Scientific Research, Kuwait
b
Civil, Environmental, and Architectural Engineering Department, University of Colorado, Boulder, CO 80309, United States

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents the results of a series of parametric analysis to investigate the factors that affect the
Received 16 July 2010 effectiveness of using simultaneously building thermal capacitance and ice storage system to reduce total
Received in revised form 31 January 2012 operating costs (including energy and demand costs) while maintaining adequate occupant comfort con-
Accepted 2 February 2012
ditions in buildings. The analysis is based on a validated model-based simulation environment and
Available online 6 April 2012
includes several parameters including the optimization cost function, base chiller size, and ice storage
tank capacity, and weather conditions. It found that the combined use of building thermal mass and
Keywords:
active thermal energy storage system can save up to 40% of the total energy costs when integrated opti-
Ice storage tank
Optimal control
mal control are considered to operate commercial buildings.
Pre-cooling Ó 2012 Elsevier Ltd. All rights reserved.
Thermal mass
Thermal energy storage

1. Introduction water) by operating the chiller during low electrical charge peri-
ods. During the on-peak periods, the ice storage is discharged to
There are two common approaches to store thermal energy in meet the building cooling requirements. As a result, it is possible
buildings. The most common approach is to install ice or chilled to reduce or even eliminate the chiller operation during on-peak
water storage tanks that are charged at night and discharged dur- hours [14].
ing the day [1–6]. An alternative approach is to utilize the thermal The parametric analysis presented in this paper is based on a
mass of the structural building materials to pre-cool the building at validated simulation environment described in a companion paper
night when the electrical rates are low. By pre-cooling the building [15]. The simulation environment is model-based and can be used
during the night and early morning hours, the thermal capacitance to evaluate the effectiveness of conventional and optimal control
of the building can be utilized to shift some of the cooling loads strategies of using both building thermal mass and ice storage sys-
from on-peak to off-peak utility rate periods [7–12]. Both ap- tem in order to reduce the operating cost while meeting the build-
proaches for storing thermal energy attempt to reduce peak cool- ing cooling requirements. The analysis includes several parameters
ing demand during the day by operating the cooling system including the optimization cost function, base chiller size, and ice
during the night. Generally, thermal energy storage (TES) systems storage tank capacity, and weather conditions.
are designed to produce the necessary cooling storage during off- First, the simulation model analysis is described including the
peak hours in order to take advantage of cheaper electric utility office building model, the cooling plant, the optimization cost func-
rates. tion, and the utility rate used in the analysis. Then, selected results
This paper investigates the effects of various design and operat- of the parametric analysis are presented and discussed.
ing factors on the optimal controls of using simultaneously build-
ing thermal capacitance and ice storage system to reduce the
2. Simulation model description
operating costs while maintaining adequate occupant comfort con-
ditions in commercial buildings [13]. The building TES can be con-
2.1. Building model
trolled by setting zone temperatures without affecting the thermal
comfort level of the occupants. The basic operating strategy of the
An office space is considered throughout the parametric analy-
active TES system is to charge the ice storage (i.e. to freeze the
ses presented in this paper. The selected office building model has
a rectangular shape with a dimension of 200 ft (61 m) by 100 ft
⇑ Corresponding author. Tel.: +1 303 492 3389; fax: +1 303 492 7317. (30.5 m). The 200 ft (61 m) walls have an east and west orientation
E-mail address: [email protected] (M. Krarti). while the 100 ft (30.5 m) walls have a north and south orientation.

0196-8904/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.enconman.2012.02.020
510 A. Hajiah, M. Krarti / Energy Conversion and Management 64 (2012) 509–515

N outlines the characteristics of the utility price structure used in

the parametric analysis presented in this paper.
West Zone
2.3. Cost function for optimal controller

South North
Using the simulation environment described in a companion
Zone Core Zone Zone 100’
paper [15], several parametric analyses are carried out to evaluate
the effectiveness of using both building thermal mass and ice stor-
age system to reduce the operating cost while meeting the building
East Zone cooling requirements. For this study, three separate optimal con-
trol strategies are considered to minimize one of three cost
functions:
200’

Fig. 1. The building model with five zones.  Energy charges only.
 Demand charges only.
 Total charges including both energy and demand charges.

Table 1 These three optimization strategies are evaluated against base-

Construction details and R-values of the building envelope. case controls:
Structure description R-value h.ft2 °F/Btu
(m2 °C/W)  Conventional control of cooling system using a fixed temper-
Roof (50 0 insulation, with 20 0 h.w. concrete deck) 18 (3.2) ature set point of 76 °F from 8:00 am to 5:00 pm This base
Exterior walls (120 0 h.w. concrete, 20 0 insulation) 9 (1.6) case is selected to investigate the effectiveness of using build-
Interior walls (40 0 h.w. concrete block with 3=4 0 0 2.4 (0.4) ing thermal mass [15].
plaster)
 Chiller-priority control when an ice storage system is utilized
Windows with light colored venetian blinds 1.2 (0.2)
[3,15].

Several parametric analyses are carried using the office building

Table 2
model to determine the effectiveness of thermal mass/ice storage
Characteristics of the utility rate structure. optimal controller under various design and operating conditions.
Selected results are presented in this paper. In this paper, selected
Rate feature Value
results obtained when the office building is located in Chicago, IL
On-peak hours (10 am–10 pm) are discussed. Typically, results for both design day (July 21) as
On-peak energy charges ($/kW h) 0.0389
well as typical day (September 28) are discussed. It should be that
Off-peak energy charges ($/kW h) 0.0208
On-peak demand charges ($/kW) 7.50 the percent cost savings obtained for typical day are indicative of
Off-peak demand charges ($/kW) 0 the annual percent cost savings associated with various control
strategies.

The height of the walls is 9 ft (2.75 m). The building model includes 3. Discussion of selected results
a core zone and four perimeter zones as illustrated in Fig. 1. The
building is occupied from 8:00 am to 5:00 pm with a density of The optimal controls for building thermal mass/ice storage pro-
100 ft2 (9.3 m2) per person in the perimeter zones and 200 ft2 per vide specific guidelines on how to operate the cooling equipment
(18.6 m2) person in the core zone. A ventilation rate of 20 cfm for 24 h period. Indeed, the optimization algorithm investigates
(9.4 L/s) per person was assumed. The construction details as well all the pre-cooling options and determines the specific hour at
as the R-values used for the walls, roof, and windows are listed in which pre-cooling starts, the length of pre-cooling, and specific
Table 1. times when the base and ice chiller need to be turned on and off.
Fluorescent lighting fixtures are used in the building model The 24 h optimization program uses initial values for the thermo-
with a power density of 1.0 W/ft2 (10.7 W/m2) operating from stat temperature settings and the ice storage state of charges. The
8:00 am to 5:00 pm daily. Moreover, equipment and appliances optimization algorithm starts its search at the specified initial con-
are modeled with a power density of 1.5 W/ft2 (16.1 W/m2) oper- ditions to find the optimized settings of the thermostat and states
ating from 8:00 am to 5:00 pm along 0.5 W/ft2 (5.4 W/m2) operat- of charge of the ice storage system that minimize the specific cost
ing continuously 24 h per day. function, i.e. energy charges, demand charges, or total energy and
demand charges. The thermal mass/ice storage optimization thus
produces a set of 24 temperature thermostat settings and a set of
2.2. Utility rate structure 24 state-of-charge levels.
The medium mass office building model described above is used
Time of use rate structure is used for the analysis of the one- in the analysis. Specifically, the simulation analysis presented in
zone building model [9,16]. Specifically, the utility rate assumes this paper uses a 30 ton base chiller, a 15 ton ice chiller, and a
the on-peak period to be from 10:00 am to 10:00 pm with off-peak 180 ton/h ice storage tank. The optimization analysis is performed
period covering the rest of the hours. On-peak energy charges are using a time-of-use utility rate structure with strong incentives for
assumed to be $0.0389/kW h while the off-peak energy charges peak load shifting as outlined in Table 2.
are $0.0208/kW h. For the on-peak demand charges, a rate of The results of the optimization analysis show that additional
$7.50/kW is used and no charges are applicable for the off-peak savings occur when building thermal mass and ice storage system
hours. This first rate is considered to have strong incentives for are used simultaneously. A brief discussion of the results obtained
cooling load shifting from on-peak to off-peak hours. Table 2 from the optimization analysis when both building thermal mass
 A. Hajiah, M. Krarti / Energy Conversion and Management 64 (2012) 509–515 511

does not reach its upper bound too early in the day. In the after-
noon, when the heat gain is at its highest level, the indoor temper-
ature set point is allowed to increase and the building thermal
mass can still absorb heat, resulting in a reduction in the cooling
load. Thus, a high demand charge forces the optimal control to
make the electric demand curve more uniform.
When optimization is based on energy charges, savings of 25.4%
in the daily energy cost relative to the base case conventional con-
trol strategy can be achieved. For demand charges optimization,
savings of 40.4% in the peak electrical demand are obtained. Final-
ly, the savings obtained when minimizing both energy and de-
mand charges are 28.5%. The results of the cooling plant daily
operating cost for the three optimized cases: optimization of en-
ergy cost only, optimization of demand cost only, and the optimi-
zation of total cost (energy and demand charges) are summarized
in Table 3. The base case consists of direct cooling from 8:00 am to
5:00 pm with no pre-cooling and no utilization of the ice storage
system. The results for the chiller priority control without pre-
cooling have been included in Table 3 for comparison purposes
Fig. 2. Indoor temperature settings obtained for the optimization of energy, to assess the benefits of using both types of thermal energy storage
demand, and total daily operating cost and for chiller priority control. systems (i.e. building thermal mass and ice storage) relative to a
conventional control strategy for ice storage systems. More de-
tailed comparative analysis of the performance of optimal controls
and ice storage system are utilized is provided in the following sec- relative to chiller priority controls is provided in the following sec-
tions for various operating and design conditions. tion. It should be noted that the chiller (50 ton) of the base case is
sized to meet the building peak design load through direct cooling.
3.1. Impact of cost function However, a 30 ton base chiller, a 15 ton ice chiller and 180 ton/h
ice storage tank are used in the chiller priority control strategy.
In this analysis, the impact of the optimization cost function on
the control strategy to operate both TES systems is investigated. 3.2. Comparative analysis of optimal controls and conventional
The indoor temperature settings for the cases when the optimiza- controls
tion is based on minimizing energy cost only, demand cost only,
and total cost (including both energy and demand charges) as well In this section, the performance of the optimal controller to
as for the baseline of using chiller priority control without any pre- minimize the total daily operating cost (including energy and de-
cooling are indicated in Fig. 2 [15]. mand charges) is evaluated against the performance incurred
For the energy cost minimization problem, the optimal control when no pre-cooling is considered but using a chiller priority con-
tends to set the building temperature at its upper bound of 76 °F trol for the ice storage system. The indoor temperature profiles
(24.4 °C) during the beginning of the on-peak period (i.e. 9:00 within the office space are illustrated in Fig. 2 for both optimal con-
am). For the demand minimization problem, the occupied on-peak trol and chiller priority control. The hourly variation of the total
period begins with the indoor temperature set at the lower bound sensible cooling load (heat extraction) for both the optimal control
of the acceptable comfort range [i.e. 68 °F (20 °C)]. and the chiller priority control are provided in Fig. 3.
When the optimization is based on minimizing both energy and It can be noticed from Fig. 3 that no heat needs to be extracted
demand charges, the optimization provides settings similar to from 5:00 pm to 9:00 pm (which marks the beginning of the unoc-
those obtained for the case of minimization of the energy charges. cupied on-peak period) when optimal control is used. Thus, the
The on-peak period starts with the indoor temperature set at the cooling system is not operated during these unoccupied on-peak
upper bound of the acceptable thermal comfort range [i.e. 76 °F hours. Indeed, the extraction of heat starts at the first hour of the
(24.4 °C)]. off-peak period (9:00 pm) using the base chiller in order to pre-
When the optimization is based minimizing the demand cost cool the thermal mass of the building. The pre-cooling continues
only, the optimization tends to have the thermostat setting at its throughout the night until the beginning of the on-peak period
lowest limits [i.e. 68 °F (20 °C)] during the first hours of occupancy. (9:00 am). From 9:00 am to 5:00 pm cooling is supplied to the
Then, the indoor temperature rises slowly during the warming-up building (i.e. heat is extracted) using both the base chiller (direct
phase so that the temperature upper bound [i.e. 76 °F (24.4 °C)] is cooling) and the ice storage system (operated in a discharging
not reached until the last hours of occupancy (3:00 pm–5:00 pm). mode).
In order to ensure the slow temperature rise, cooling is provided The time variation of the ice storage state of charge for both
throughout the occupancy period. To prevent a high peak demand optimal control and chiller priority control is displayed in Fig. 4.
in the cooling load, the indoor temperature rise is delayed so it For the optimal control, the ice storage charging process (using

Table 3
Performance of three optimal controls for minimizing energy, demand, and total daily operating cost relative to conventional and chiller-priority controls.

Energy cost ($) Demand cost ($) Total cost ($) Savings in total cost (%)
Base case 106.1 34.4 140.4 –
Chiller priority 96.5 29.8 126.3 10.0
Minimizing energy cost 79.1 21.3 100.4 28.5
Minimizing demand cost 97.5 20.5 117.9 16.0
Minimizing total cost 79.1 21.3 100.4 28.5
512 A. Hajiah, M. Krarti / Energy Conversion and Management 64 (2012) 509–515

Fig. 5. Total cooling electrical power (kW) obtained by optimization of the total
Fig. 3. Heat extraction (ER) obtained for optimal control and for chiller priority daily operating cost compared to that incurred using chiller priority control.
control.

Fig. 4. Ice storage state of charge obtained by optimization of the total daily
Fig. 6. Base chiller electrical power (kW) obtained by optimization of the total
operating cost compared to that incurred using chiller priority control.
(both energy and demand) operating charges compared to that incurred using
chiller priority control.

the ice chiller) starts immediately at the first off-peak hour (9:00
pm). The ice chiller continues to charge the ice tank until the end
of the off-peak period (9:00 am) when 88% of the tank capacity
is charged (x = 0.88). The ice storage discharge process starts at
9:00 am and continues until the end of occupancy (5:00 pm). In-
deed, the ice tank is progressively depleted during the period span-
ning from 9:00 am to 5:00 pm to assist the base chiller.
The total cooling electrical power (kW) for the optimal control
and the chiller priority control is presented in Fig. 5. The electrical
power for base chiller and ice chiller are shown in Figs. 6 and 7,
respectively.

3.3. Sequential optimal controls

In this section, the benefits of using integrated optimization to

develop optimal controls that are capable of operating simulta-
neously passive and active thermal energy storage (TES) systems
are assessed. In the integrated optimization, 48 variables (24 tem-
perature settings and 24 charging/discharging rates) are optimized
for each day as discussed in [15]. First the benefits of the integrated Fig. 7. Ice chiller electrical power obtained by optimization of the total operating
optimization of both TES systems are evaluated compared to using charges compared to that incurred using chiller priority control.
 A. Hajiah, M. Krarti / Energy Conversion and Management 64 (2012) 509–515 513

Table 4 clusion is valid for other building types, climate conditions, and
Cost savings obtained for optimal TES control, optimized pre-cooling, and optimal utility structures.
control using both building thermal mass and ice storage system.

Control type Energy Demand Total Savings in 3.4. Impact of design parameters
cost ($) cost ($) cost ($) total cost (%)
Conventional (no TES 106.1 34.4 140.4 – This section investigates the effects of cooling system capacities
system) on the total operating cost of the cooling system using the devel-
Optimized pre-cooling 91.3 28.2 119.5 14.9
Optimized ice storage 96.0 29.3 125.3 10.8
oped integrated TES optimal controller. Specifically, the impacts
system of the base chiller size, ice chiller size, and ice tank size are evalu-
Optimized pre-cooling and 81.0 20.0 101.0 28.1 ated under design day conditions and typical day conditions.
ice storage system
3.5. Effect of the base chiller size

3.5.1. Base chiller size effect under design conditions

Table 5 Table 6 summarizes the results of an analysis that investigates
Comparison of the performance of sequential optimization, integrated optimization, the effect of varying the size of the base chiller on the total (energy
and optimized pre-cooling in minimizing the daily total cost. and demand) operating cost of the cooling system for the design
Control type Energy Demand Total Savings in day. The results provided in Table 6 indicate that optimal controls
cost ($) cost ($) cost ($) total cost (%) can achieve cost savings in the range of 22–28% relative to the con-
Conventional (no TES 106.1 34.4 140.4 – ventional operating strategy. The cost savings are the highest for
system) the design day conditions when the base chiller size is 30 ton.
Optimized pre-cooling 91.3 28.2 119.5 14.9 For base chiller capacities lower than 30 ton, the cost savings
Optimized ice storage 81.8 20.9 102.7 26.9 decrease significantly as the base chiller capacity is reduced. This
system
Optimized pre-cooling and 81.0 20.0 101.0 28.1
behavior is due to the fact that a small base chiller does not provide
ice storage system enough cooling capacity to pre-cool the building thermal mass to
desirable low temperatures. Meanwhile, when the base chiller is
higher than 30 ton, the cost savings is slightly reduced (relative
to the savings of 28.5% achieved for a 30 ton base chiller) but re-
only one TES system at a time. Table 4 summarizes the results of mains at the same level of 28.1% for any large base chiller. This re-
using optimal ice storage system control, optimized pre-cooling, sult is due to the part-load performance of the chiller. Indeed,
and optimal control using both building thermal mass and ice stor- when the base chiller is large, the cooling loads for both pre-cool-
age system for the office building (medium mass) located in Chi- ing and direct cooling represent smaller fractions (relative to the
cago, IL. case of 30 ton base chiller) of the chiller capacity. Thus, the larger
The ice storage system optimal controller achieved total cost base chillers have to be operated at lower part load ratios and thus
savings of 10.8% relative to the conventional control (system oper- operate less efficiently.
ating during occupancy hours only with no pre-cooling). However,
higher savings of 28.1% in the total daily operating cost are ob- 3.5.2. Base chiller size effect under typical conditions
tained with the use of the combined optimal controller (i.e. use A similar analysis that investigates the effect of varying the size
of both pre-cooling and ice storage) relative to the conventional of the base chiller on the total (energy and demand) operating cost
control. This result illustrates clearly the advantage of using both of the cooling system has been performed for the typical day (i.e.
passive thermal energy storage system (i.e. building thermal mass) the 28th day of September in Chicago, IL). The savings in the total
and an active thermal energy storage system (i.e. ice storage tank) cost were identical for the three base chiller capacities (50, 25, and
to minimize total building energy costs.
Instead of the integrated optimization (with 48 variables/day to
Table 6
identify), sequential optimization is considered to reduce the com-
Effect of the base chiller size on the total daily cost for a design day.
putation efforts. The sequential optimization is based on using
optimized pre-cooling first followed by an optimized use of the Base chiller (ton) Total daily cost ($) % Savings
(relative to base case)
ice storage system to minimize the cost function (i.e. energy cost,
demand cost, or total cost). Thus, the sequential optimization iden- 50 (base case) 140.40 –
tifies only 24 variables per day at a time. Specifically, the hourly 20 108.40 22.80
25 103.0 26.60
thermostat settings for the space temperatures resulted from the 30 100.40 28.50
optimized pre-cooling of building thermal mass are used as inputs 35 100.50 28.40
in the TES optimization (i.e. the charging/discharging of ice storage 40 101.0 28.10
tank). The results of a comparative analysis for minimizing the to- 45 101.0 28.10
50 101.0 28.10
tal cost using both sequential and integrated optimization schemes
are shown in Table 5.
As indicated in Table 5, the sequential optimization provides al-
most similar cost savings than the integrated optimization at least
Table 7
for the building, HVAC system, and utility rate structure considered Effect of the base chiller size on the total daily cost for a design day.
in the analysis. Thus, sequential optimization (i.e. optimization of
Base chiller (ton) Total daily % Savings
pre-cooling strategy that utilizes the building thermal mass fol-
cost ($) (relative to base case)
lowed by optimization of the charging/discharging of ice storage
tank) can be considered instead of a more complex integrated opti- 50 (base case) 121.0 –
15 93.3 22.9
mization that attempts to simultaneously determine the optimal 25 93.7 22.6
settings for both pre-cooling and ice storage system operation. 50 94.4 22.0
However, further investigation is needed to determine if this con-
514 A. Hajiah, M. Krarti / Energy Conversion and Management 64 (2012) 509–515

Table 8 Table 10
Effects of the ice chiller size on the total daily cost for design day. Effect of the ice tank capacity on the total daily cost for design day.

Ice chiller (ton) Total daily cost ($) % Savings (relative to base case) Ice tank (ton h) Total daily cost ($) % Savings (relative to base case)
Base case 140.40 – – 140.40 –
5 117.40 16.40 75 114.10 18.73
10 110.70 21.20 100 112.10 20.20
15 103.0 26.64 125 109.90 21.72
20 102.20 27.21 150 107.90 23.15
25 101.90 27.42 175 104.30 25.7
200 101.70 27.60

15 tons) selected in the analysis. Table 7 summarizes the results of

Table 11
this analysis that investigates the effect of varying the size of the Effect of the ice tank capacity on the total daily cost for design day.
base chiller on the total (energy and demand) operating cost of
the cooling system under typical day conditions. Ice tank (ton h) Total daily cost ($) % Savings (relative to base case)
– 121.0 –
60 99.9 17.4
3.6. Effect of the ice chiller size 100 93.9 22.4
180 93.3 22.9

3.6.1. Ice chiller size effect under design conditions

Table 8 summarizes the results of an analysis that investigates
the effect of varying the size of the ice chiller on the total operating of 10 ton. Any larger ice chiller capacity does not provide signifi-
cost of the cooling system for design day conditions. The ice chiller cant additional load shifting potential.
of the cooling plant is designed only to charge the ice tank during
off-peak hours. It cannot be used for direct cooling during occupied 3.7. Effect of the ice tank size
hours or for pre-cooling during unoccupied hours and off-peak
hours. 3.7.1. Ice tank size effect under design conditions
As indicated in Table 8, the cost savings achieved by optimal Table 10 summarizes the results of an analysis that determines
control increase with the size of the ice chiller. The rate of increase the effect of varying the size of the ice tank on the total operating
becomes significantly small when the ice chiller capacity is above cost of the cooling system under design day conditions. Table 10
15 ton. These results stem from the fact that for small ice chiller also shows the percent cost savings in total electrical energy cost
capacities, the load-shifting benefits of ice storage systems are obtained by using optimal controls relative to the conventional
not fully realized. These benefits are almost completely achieved operating strategy (i.e. direct cooling without pre-cooling or use
for an ice chiller capacity of 15 ton. Any larger ice chiller capacity of TES system).
does not provide significant additional load shifting potential. As expected, the results of Table 10 indicate the cost savings po-
tential incurred from optimal control increases with the size of the
ice storage tank. Indeed, the higher the ice storage system, more
3.6.2. Ice chiller size effect under typical conditions
load-shifting can be achieved from on-peak to off-peak hours.
Table 9 summarizes the results of an analysis that investigates
the effect of varying the size of the ice chiller on the total operating
cost of the cooling system for typical day conditions. The results 3.7.2. Ice tank size effect under typical conditions
provided in Table 9 indicate that optimal controls can achieve cost Table 11 summarizes the results of an analysis that determines
savings in the range of 17–23% relative to the conventional operat- the effect of varying the size of the ice tank on the total operating
ing strategy (i.e. system operating from 8:00 am to 5:00 pm to pro- cost of the cooling system under typical day conditions.
vide direct cooling). The cost savings are the highest for the typical Similar to the findings with design day conditions, the results of
day conditions when the ice chiller size is about 10–15 tons. Table 11 indicate the cost savings potential incurred from optimal
Similar to the findings for design day conditions, the cost sav- control increases with the size of the ice storage tank. The results
ings achieved by optimal control increase with the size of the ice indicate that optimal controls can achieve cost savings in the range
chiller. The rate of increase becomes significantly small when the of 17–23% relative to the conventional operating strategy (i.e. base-
ice chiller capacity is above 10 ton. These results stem from the fact case). The cost savings are the highest for the typical day condi-
that for small ice chiller capacities, the load-shifting benefits of ice tions when the ice chiller size is about 180 ton h. Thus, large ice
storage systems are not fully realized. With the 5 ton ice chiller, tanks provide more total cost savings for both design and typical
even though the charging period of the ice tank is longer than days. Economical analysis should be carried out to estimate the
the cases with larger capacities, the maximum state of charge of cost-effective ice tank size.
the ice tank is 50%. Moreover, the base chiller is needed to meet
the building load during the occupancy period and is operating 4. Summary and conclusions
at a low part load ratio. The load-shifting benefits of ice storage
systems are almost completely achieved for an ice chiller capacity This paper has investigated some of the important factors that
affect the performance of optimal controls of using simultaneously
building thermal capacitance and ice storage system to reduce the
Table 9
cooling system total operating costs (including energy and demand
Effect of the ice chiller size on the total daily cost for typical day.
costs) while maintaining adequate occupant comfort level condi-
Ice chiller (ton) Total daily cost ($) % Savings (relative to base case) tions in office buildings. The building thermal storage is utilized
Base case 121.0 – through pre-cooling strategies by setting space temperatures to
5 100.0 17.4 specific values during unoccupied and occupied hours. Moreover,
10 93.4 22.8
an ice storage tank is charged by operating the chiller during low
15 93.3 22.9
electrical charge periods. During on-peak periods, the ice storage
 A. Hajiah, M. Krarti / Energy Conversion and Management 64 (2012) 509–515 515

is discharged to meet the building cooling requirements. As a re- [3] Henze GP, Krarti M, Brandemuehl MJ. A simulation environment for the
analysis of ice storage controls. Int J HVAC&R Res 1997;3(2):128–48.
sult, it is possible to reduce or even eliminate the chiller operation
[4] Massie DD. Optimal neural-network-based controller for ice storage systems.
during on-peak hours. PhD dissertation. University of Colorado, CO; 2008.
The optimization results discussed in this paper for a typical of- [5] Henze GP, Krarti M. The impact of forecasting uncertainty on the performance
fice building model under different weather conditions and various of a predictive optimal controller for thermal energy storage systems. ASHRAE
Trans 1999;105(2):553–61.
design options indicate that significant cost savings (up to 40%) can [6] Braun JE. A near-optimal control strategy for cool storage systems with
be achieved in the cooling system total operating cost. Generally, dynamic electric rates. Int J HVAC&R Res 2007;13(4):557–80.
the results indicate that optimal control for both building thermal [7] Keeney KR, Braun JE. A simplified method for determining optimal cooling
control strategies for thermal storage in building mass. Int J HVAC&R Res
mass pre-cooling and ice storage operation outperforms all of 1996;2(1):59–78.
other conventional controls and sequential optimal controls under [8] Morgan S, Krarti M. Impact of electricity rate structures on energy cost savings
all climate conditions, utility rate structures, and system designs. of pre-cooling controls for office buildings. Build Environ 2006;42(8):2810–8.
[9] Henze GP, Felsmann C, Florita AR, Brandemuehl MJ, Cheng H, Waters CE.
However, the analysis presented in this paper showed that sequen- Optimization of building thermal mass control in the presence of energy and
tial optimal control can achieve the majority of the cost savings po- demand charges (1313-RP). ASHRAE Trans 2008;114(2):75–84.
tential of the developed optimal controls for both pre-cooling and [10] Henze GP, Florita AR, Brandemuehl MJ, Felsmann C, Cheng H. Advances in
near-optimal control of passive building thermal storage. In: ASME 3rd
ice storage charging/discharging. international conference on energy sustainability. San Francisco (CA); 2009.
While laboratory testing have been carried to validate the cost [11] Sebzali MJ, Rubini PA. Analysis of ice cool thermal storage for a clinic building
savings benefits of the optimal controls using both building mass in Kuwait. Energy Convers Manage J 2006;47(19):3417–34.
[12] Qureshi WA, Nair N-KC, Farid MM. Impact of energy storage in buildings on
and ice storage, additional fields testing of the developed optimal
electricity demand side management. Energy Convers Manage J
controls would be the following natural step in further investigat- 2011;52(5):2110–20.
ing the benefits of optimal controls for using simultaneously build- [13] Kintner-Meyer M, Emery AF. Optimal control of an HVAC system using cold
ing thermal mass and thermal energy storage system. storage and building thermal capacitance. Energy Build 1995;23(1):19–31.
[14] Liu S, Henze GP. Experimental analysis of simulated reinforcement learning
control for active and passive building thermal storage inventory – part I:
References theoretical foundation. Energy Build 2006;38(2):142–7.
[15] Hajiah A, Krarti M. Optimal control of building storage systems using both ice
[1] Drees KH, Braun JE. Development and evaluation of a rule-based control storage and thermal mass, part I: simulation environment. J Energy Manage
strategy for ice storage systems. Int J HVAC&R Res 1996;2(4):312–36. Convers 2011.
[2] Henze GP, Krarti M. Ice storage system controls for the reduction of operating [16] Energy and environmental economics. A survey of time-of-use pricing and
cost and energy use. ASME Sol Energy Eng J 1998;120(4):275–82. demand response programs. EPA Report. Washington (DC); 2006.