Energy Conversion & Management 40 (1999) 15271542

Estimation and measurement of electric energy

consumption due to air conditioning cooling load
Omar M. Al-Rabghi*, Mohammed H. Al-Beirutty, Kadry A. Fathalah
Thermal Engineering Department, King Abdulaziz University, P.O. Box 9027, Jeddah 21413, Saudi Arabia
Received 23 June 1998; accepted 22 January 1999

Abstract

Electric energy consumption due to air conditioning (A/C) in buildings is tremendous, especially in
places having hot humid weather, such as Jeddah, Saudi Arabia. Building simulation programs, such as
DOE2, are very valuable tools to study building energy consumption under dierent conditions. These
programs can be used to investigate the eect of dierent building designs on their energy consumption.
This paper presents the experimental and theoretical variations of electric power consumed by an A/C
system. For the experimental part, a building has been chosen, and dierent sensors were installed to
measure and record electric kW consumption at dierent indoor and outdoor conditions. In the
theoretical part, the Visual DOE simulation program, which is based on the well known DOE2E
program, has been used. Actual building data was fed to the program to generate the instantaneous
electric consumption due to A/C cooling load. Comparison between measured and predicted electric
consumptions is presented. # 1999 Elsevier Science Ltd. All rights reserved.

1. Introduction

Air conditioning (A/C) in buildings is no more a luxurious request, rather it is a top priority
and need. Although there are dierent means by which refrigeration (i.e. cooling eect) is
produced, the national most common method to provide comfort condition in a building in
Saudi Arabia is by vapor compression. The heart of the vapor compression cycle is the
compressor which, in most cases, needs electric energy to run. Electric energy supply is widely
available, making the vapor compression cycle the easiest and most preferred refrigeration

* Corresponding author. Tel.: +966-2-640-2000; fax: +966-2-965-2182.

E-mail address: [email protected] (O.M. Al-Rabghi)

0196-8904/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 9 6 - 8 9 0 4 ( 9 9 ) 0 0 0 4 5 - X
1528 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542

method of choice. This is the reason why the demand for electricity is high in the Kingdom,
especially during the summer months. Over 60% of the produced electric energy in a large city
like Jeddah, Saudi Arabia in summer is consumed by A/C. The main function of an A/C
system is to overcome the cooling load of a building while providing comfort to its occupants.
Cooling load is dened as the sum of all loads (internal and external) to be removed from a
space so as to keep its condition (mainly temperature) constant. Cooling load is actually the
sum of the heat gains of the building that get transferred to the space air. There are dierent
types of heat gains, like solar, heat transmission through walls, human, lights, ventilation and
inltration. Depending on the building mass, capacity and construction, these heat gains are
converted to cooling loads after some time delay. The heat extraction rate is dened as the
actual heat removal rate from the space by the A/C system. If the space air temperature is to
be maintained at a constant value, the heat extraction rate must be equal to the cooling load.
In practice, however, due to A/C control, the space air temperature experiences a swing, i.e. a
temperature uctuation. Both the cooling load and the heat extraction rate are functions of the
outside and inside building conditions.
The actual load on the central refrigeration plant (for example, a chiller) equals the heat
extraction rate plus the A/C load. The A/C load includes the loads due to duct friction, fan,
losses etc. Obviously, the plant capacity (chiller, for example) changes with ambient conditions
and system parameters, such as chilled water ow rate. The consumed electric power due to A/
C, is due to the mechanical work needed to operate the central plant (compressors, pumps,
etc.) plus any additional electric work needed to provide comfort conditioning, like the work
for fans, pumps, control system etc.
The demand for electricity is mostly related to the ambient conditions, especially the ambient
temperature. A look at the consumed kW curve for Jeddah during the summer reveals such a
relation [8]. The maximum electricity demand is around 3 p.m.
Because of building construction expansion in the Kingdom and the increasing demand for
comfort conditioning, the electric companies experience an increasing demand for electric
power. The yearly growth of this demand is among the highest world wide [8]. The Ministry of
Industry and Electricity, along with the electric companies in the Kingdom, have prepared
electric saving programs to reduce the demands or at least shift the time of the peak demand.
Each of the four electric companies in the Kingdom has facilitated its own method for
lowering the electric energy consumption [8]. Some of these tried methods were successful in
curving down the electric demand [8].
In this study, the electric kW consumed by an A/C system is measured and recorded to see
the time variation of this parameter. Indoor and outdoor air conditions and solar intensities
are also measured. The same parameters will also be predicted by a simulation program called
Visual DOE2-5 [4]. This graphic, easy to use computer program, is based on the famous DOE-
2E simulation engine [6].

2. Building description

The Solar Energy Building which belongs to the Thermal Engineering Department has been
selected for the study. A three-dimensional (3D) view of the building is shown in Fig. 1. The
 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542 1529

Fig. 1. 3D view of the selected building under study.

building consists of ground unconditioned space and an air-conditioned rst oor. Because of
the existence of the solar collector shed and building C, see Fig. 1, only the east and north
sides of the building are exposed to direct sun radiation. A large portion of the roof is shaded
by ve long line collectors, chillers and other equipment.
The A/C system consists of two identical air cooled chillers, each of capacity 42 ton. Each
chiller consists of two compressors. One is of 30 hp capacity and the other is of 20 hp capacity.
The compressors are reciprocating and semi-hermetic. Each condenser has six fans with a 1 hp
motor for each fan. There are 32 fan coil units located in dierent rooms in the rst oor of
the building. The chilled water is circulated to these fan coil units by a 5 kW motor pump.
Since the air is conditioned by these ceiling fan coil units, fresh air is provided to dierent
spaces by a special fresh air unit. This unit is placed on the roof. It has a fan motor rated at 3
hp. Each room at the rst oor has a thermostat and a fan switch. The thermostat can be set
to the desired temperature, and the fan switch can vary the fan speed (i.e. Low, Medium,
1530 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542

Fig. 2. Plan view of the rst oor at solar energy building (dimensions in m).

High, O). According to design, there exists a solenoid valve on each fan coil which is
controlled by the room thermostat to control the water ow rate (ono) into the coil. Fig. 2
shows a plan view of the rst oor under study.

Fig. 3. Measured A/C electric power consumption during October 2225, 1997.
 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542 1531

3. Instrumentation and measurements

Two PCs were used to collect the experimental measured data. One PC is used mainly to
measure electric power consumption due to the A/C cooling load. This PC is kept in the
electric room near the A/C electric power panel. The other PC was used to measure and collect
indoor and outdoor air-conditions data, in addition to solar intensities. A PC board from
Highland Technology, model AT-180 was used [5]. This board is capable of continuously
measuring and recording electric power parameters, such as instantaneous electric variables
(e.g., RMS (root mean square) current, RMS voltage, phase angle), and instantaneous power
in kW and in kVA. Integrated power, i.e. the energy consumed in kWh, is also calculated. In
addition, the board is capable of receiving analog, resistive and digital signals. Both single and
multi-phase electric arrangements can be handled. The data measured can be stored on hard or
oppy computer disk, on time intervals selected by the user. In order to measure the power,
the current and voltage are measured continuously. Currents are measured using current
transducers (CT) which measure current up to 400 A. The milli-volts reading of the CT is fed
to the board through a terminal panel. Usually, one CT is used for each phase. Voltage up to
500 V is fed to a special voltage transducer, where the signal is stepped down to milli-volts,
then fed to the board through the terminal panel. For more details about the AT-180, its
capabilities, CT, voltage transducer and AT-180 programming, the reader is referred to
Highland Technology publications [5].

Fig. 4. Measured solar radiation intensities during October 2426, 1997.

1532 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542

Before using the AT-180 board to measure the actual A/C electric power, evaluation and
check tests were performed to double check the measured values. The AT-180 readings were
found to be very accurate.
There exists a separate panel just for the A/C system. From this electric panel, electricity is
provided to the chillers, pumps, fan coil units etc. (i.e. all A/C equipment). Adjacent to this A/
C panel, a smaller panel just for the 32 fan coil units is located. 3 CTs were installed on the
main A/C panel to measure the total kW consumed by the whole A/C system. Also, we have
installed three CTs on the fan coil unit panel to measure the power consumed by the fan coil
units. The extra feature of the AT-180, which is receiving analog signals, is used to measure
and record indoor air conditions. Three rooms on the rst oor were selected to measure their
temperatures and relative humidities.

3.1. Data acquisition for indoor and outdoor conditions

The following inside and outside parameters were measured and recorded by the second PC:

(a) Outdoor air temperature and relative humidity.

(b) Indoor air temperature and relative humidity. The indoor temperature was measured at
several locations.
(c) Solar radiation intensities.

Fig. 5. Solar radiation intensities as predicted by ASHRAE clear sky model during the period October 2426, 1997.
 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542 1533

This PC was kept in the solar energy laboratory. The above parameters were recorded each 15
min, which is the same time interval as for the rst PC.

4. Experimental results

Fig. 3 shows the total measured A/C power consumption as a function of time. The gure
depicts the total electric power consumed by the A/C equipment for approximately 5 days
(October 2225, 1997). As seen in the gure, the electric kW varies around 65 kW. The
dierence between the maximum and minimum kW is roughly 5 kW, which is believed due to
variation of the condenser pressure due to the variation in ambient temperature. Inspecting the
chiller control, the only information that could be obtained is that the two large compressors
(30 hp each) were providing the colling requirement for the building according to the demand.
It was noticed that only one of these large compressors has cylinder unloaders.
The kW consumption when both chillers are o is roughly 12 kW. This is the measured kW
electric consumption at times when the maintenance department is servicing the chillers. The 12
kW is the power for everything excluding the chillers. When one of the large compressors is
o, Fig. 3 suggests that the consumption is around 38 kW. Therefore, the electric consumption
for each of these large compressors is 26 kW, with possibilities to go down to 22 kW due to
ambient temperature variations. It is very clear from the gure that the cylinder unloaders are
not working, and the control strategy is either the two compressors are ON or OFF. The

Fig. 6. Measured outdoor air relative humidity and temperature during October 2325, 1997.
1534 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542

system trip os are believed to occur due to freezing. Actually, icing has been noticed on
several occasions. White ice was obvious on the external surface of the cooler.
As a matter of fact, the system has a large capacity in comparison with the required cooling
coil load or chiller load. The reason behind this is the small required cooling load of the
building is due to the following reasons:

1. Part of the roof is covered with solar collectros, chillers, fresh air unit and solar energy
student's projects.
2. The west side of the building is almost totally shaded due to the existence of the solar
collector shed, cooling tower and building C, see top view of the building.
3. The ground oor of the building has a negligible load on the rst oor under study due to
the fact that part of the ground oor is shaded with practically small loads. The other part
of the ground oor is utilized by the Aeronautical Department, and it is air conditioned
with totally separate A/C units.
4. Only the rst oor of the building plus the two entrances are air conditioned by these two
chillers.
5. The people and equipment loads are very small since few people are utilizing the building,
and no major heat load equipment exists in the building.

Because of the above reasons and based on a rough cooling load estimation, the decisions of

Fig. 7. Measured indoor temperature and relative humidity for solar energy building during October 2325, 1997.
 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542 1535

the A/C maintenance department to run the two large compressors is not wise. This actually
creates a very low temperature in space and causes tripping of the chiller due to freezing, as
will be seen later.
The measured solar intensities (i.e. direct, diuse and total) are shown in Fig. 4. The gure
shows the variations in intensities for 3 days in the month of October. For the sake of
comparison, the ASHRAE clear sky model for solar intensities is shown in Fig. 5. Notice that
the peak of both, the ASHRAE model [2] and the experimental results, occurs at 12 p.m.
which is approximately solar noon. The dierence between the ASHRAE and measured peaks
is around 10%.
The measured outdoor air temperature and relative humidity are shown in Fig. 6. The
variations of these two parameters are plotted for October 2325. Notice that we have tried to
plot all the parameters (i.e. A/C power, intensities, indoor air condition and outdoor air
condition) for the same time period i.e. around October 23, to make the comparison
meaningful. We could not produce all the parameter variations for the same period accurately,
because we have two PCs (i.e. two data acquisition systems) that have some time lag, in
addition to interruptions due to installation of new sensors, power failure and/or curious
intruders. Notice in Fig. 6, the expected temperature behavior, i.e. the maximum temperature
occurs around 23 p.m. and the minimum around 5 a.m. The outside relative humidity varies
opposite to the temperature, i.e. as the outside temperature increases (i.e. air heating), the
relative humidity falls. The minimum relative humidity is 23 h after solar noon. This suggests
that the humidity ratio of the outside air is almost constant, and as the air is heated, its
relative humidity decreases.

Fig. 8. Measured A/C electric power consumption during the period November 1924, 1997.
1536 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542

The measured indoor conditions (i.e. air temperature and relative humidity) are shown in
Fig. 7. The air relative humidity in the space is 60%, on average, and does not decrease below
55% or increase over 65% for the 3 days shown in the gure. Although the set point indoor
temperature is around 24258C, the measured space temperature, as shown in the gure, varies
between 19.5 and 21.58C. This reects the problem of an over sized A/C system, having a poor
control strategy. Additionally, according to design, there exists a solenoid valve at each fan coil
unit. This valve should operate according to the thermostat setting. Evidently, if not all, most
of these valves are not working satisfactorily.
After repeated complaints to the A/C maintenance department, they agreed to change the
chiller control and turn only one chiller ON. However, the results are still not encouraging.
The measured kW for the period November 1424 is shown in Fig. 8. This variation denitely
indicates that the refrigeration control has worsened, with ono cycling (most probably due
to freezing). This causes a current demand variation in the order of 100 A, cycling more than
100 times a day. This behavior causes a problem to the electric company due to frequent
variations in the in-rush currents, as well as reducing appreciably the life span of the A/C

Fig. 9. Constructed occupancy schedule for solar energy building.

 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542 1537

machine, especially the heart of the unit which is the compressor. In some countries, the
electric companies penalize consumers with high frequent variations in current.
One notices though that the indoor air temperature signicantly drops to below 238C due to
improper A/C control, see Fig. 7. This denitely causes two unwanted problems. First, it
produces unhealthy and thermally uncomfortable indoor conditions. The other problem is
running the chiller at low temperature, lowers its eciency, which leads to higher power
consumption. Additional experimental results are available in Ref. [1].

5. Visual DOE-2.5 simulation program

The DOE-2 program [6] is a set of more than 30 les and subprograms written for the sole
purpose of simulating energy use in buildings. The Department of Energy in the United States,
along with energy related utilities, have sponsored a group of researchers at LBL (Lawrence
Berkeley Laboratory) and Los Alamos Laboratories to develop the FORTRAN computer
code. The program has gone through dierent developing stages and versions. The current

Fig. 10. Total cooling load on chillers (October 2228, 1997).

1538 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542

version of the program is DOE-2E [7]. The program is heavily used by engineers, architects
and utility managers to estimate the energy requirement for buildings. The most important
aspect of DOE-2 is the facility it has to predict the cost eective design of dierent building
design alternatives.
Basically, DOE-2 program is a modeling and simulation code. It accepts building
construction data, weather data, HVAC information, equipment, lights and people occupancies
as input data. The program calculates the building hourly cooling and heating loads for the
whole year (i.e. 8760 h). These loads, along with the control strategy schedule for the HVAC,
are converted into A/C system loads. In another stage, these system loads are then manifested
on the central plant as plant loads. Knowing the type, cost of energy and scheduling, an
economic simulation is made on the plant loads to estimate the energy costs. The cost analysis
can be done for dierent alternative designs of the basic case to evaluate dierent designs or
construction parameters. The DOE-2, therefore, is essentially a collection of simulation
programs of loads, systems, plant and economics.
In addition to the famous DOE-2 simulation program, there are other simulation programs
such as BLAST and TRANSYS. BLAST stands for Building Loads Analysis and System

Fig. 11. Cooling electric kW required to over come cooling load (October 2228, 1997).
 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542 1539

Thermodynamics, which was developed at the University of Illinois [3]. TRNSYS stands for
the transient system simulation program developed at the University of Wisconsin [9]. DOE-2
is under continual revision and updating. The maintainers of the program are at LBNL
(Lawrence Berkeley National Laboratory). The new version (i.e. DOE-2.2) with added features
is expected in the near future.
Visual DOE-2.5 [4] is a powerful Windows based, graphic interface to the well known energy
analysis program DOE-2.1E. This graphic interface has enabled engineers, architects, energy
consultants and other concerned professionals to simulate quickly and predict the performance
of their energy systems and compare the savings associated with dierent design alternatives.
Because of the graphic nature of the Visual DOE interface, a building model and HVAC
system diagram are easily produced and can be visualized as they are created.
The Visual DOE-2.5 was used to build a model that simulates the Solar Energy Building at
King Abdulaziz University. Plan and elevation views were drawn based on the dimensions of
the As-Built drawings of the walls, oors and roofs. Construction type materials were built
using the Construction Builder module. A glazing construction similar to that of the actual
fenestration was selected and assigned to all glazing surfaces of the building. Dimensions of the
glazing surfaces were set as per actual data. Architectural detail and summary reports are given
in the nal technical report [1].
Occupancy schedules were constructed for people, lights, equipment, inltration, indoor
temperature settings and for HVAC system operation. Fig. 9 shows the predicted week day
occupancy of the building as a fraction of the maximum expected occupancy, which mainly

Fig. 12. Chiller performance on October 23cooling electric kW vs. cooling load on chiller.
1540 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542

would be students attending the only class room available in the building with a few
technicians occupying other spaces of the building. Similar schedules for lights, equipment and
inltration were also constructed.

6. Visual DOE results

In order to study the performance of the operating HVAC system and estimate the electrical
kW required for cooling, simulation was performed and hourly reports were obtained for a
one-week period, i.e. 2228 October. Figs. 1013 show the results obtained for the specied
period.
Fig. 10 shows the calculated building total cooling load on the chillers. The gure predicts
the same cooling load for week days with a maximum value of about 95 kW that takes place
at about 1 p.m. versus 72 kW on Thursdays and 65 kW on Fridays (Kingdom's weekend is
Thursday and Friday). The electric kW required to operate the compressors is given in Fig. 11.
A maximum of 27 kW would be required to operate the chillers plant on week days, while 22
kW are required on Thursdays and 21 kW on Fridays. A comparison of the chiller cooling
load with the required electrical kW for operation for the 23 October is shown in Fig. 12.
Fig. 13 shows the monthly electric demand by the solar energy building to operate the

Fig. 13. Monthly total electric demand of the solar energy building.
 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542 1541

Fig. 14. Monthly total electricity consumption of the solar energy building.

scheduled equipment, lights, and HVAC system. About 70 kW would be required to operate
the whole building electric facilities. The building energy consumption in kWh per month is
shown in Fig. 14. The low consumption in June and September is mainly due to the summer
vacation period which was scheduled between 18th June and 6th September. During this
period, all electric facilities are assumed to be turned o. The same comment applies to July
and August. A maximum monthly electric consumption of about 30,000 kWh is estimated in
May. The annual consumption of electric energy adds up to 240,844 kWh, out of which about
43.4% is consumed by cooling equipment and 25.8% is consumed by the HVAC.

7. Conclusions and recommendations

The following conclusions and recommendations were derived:

1. Experimental measured results for indoor and outdoor conditions are reasonable.
Experimental results for the chiller electric kW consumed, however, indicate a lot of tripping
due to impaired refrigeration system control on the chillers as well as on the A/C system.
Replacement or xing of both control systems is necessary.
1542 O.M. Al-Rabghi et al. / Energy Conversion & Management 40 (1999) 15271542

2. The Visual DOE-2.5 simulation program is a powerful, graphical, easy to use tool for
predicting energy use in buildings. Visual DOE-2.5 lacks good comprehensive
documentation.
3. The project team recommends further study on other buildings, provided that the selected
buildings have a reliable energy ecient A/C control.

Acknowledgements

This work has been supported by King Abdulaziz University through grant no. 104/1417
(1997). The rst author thanks Eng. Faisal Al-Khlaiwi for helping in carrying the experimental
part of the project as part of his B5 project.

References

[1] Al-Rabghi OM, Fathalah K, Al-Beirutty M. Final Technical Report on Estimation and Measurement of Electric
Consumption due to Air Conditioning Cooling Load in Building, King Abdulaziz University, Thermal
Engineering Department, Funded Project No. 104/1417, 1998.
[2] ASHRAE. Handbook of Fundamentals. American Society of Heating, Refrigeration and Air Conditioning
Engineers, Atlanta, Georgia, USA, 1981.
[3] BLAST (Building Loads Analysis and System Thermodynamics), 30 Mechanical Engineering Building,
University of Illinois, 1266 W. Green Street, Urbana, IL 61801, USA.
[4] Eley Associates. Visual DOE-2.5 Developers, 142 Minna Street, San Francisco, CA 94105, USA, 1996.
[5] Highland Technology Technical Manual for AT-180. Highland Technology, 320 Judah Street, San Francisco,
CA 94122, USA.
[6] LBL. DOE-2 Reference Manual Parts 1 and 2, Version 2.1A with 2.1B Supplements. Lawrence Berkeley
Scientic Laboratories, Available from NTIS, National Technical Information Service, USA, 1980.
[7] LBNL (Lawrence Berkeley National Laboratory). Simulation Research Group MS: 90-3147, University of
California, Berkeley, CA 94720, USA.
[8] SCECO. Technological conservation of electric energy consumption. A key lecture presented by the Ministry of
Industry and Electricity to the 4th Saudi Engineering Conference Jeddah, Saudi Arabia (in Arabic), 1995.
[9] TRANSYS, Transient System Simulation Program. Kleins, Beckman W, et al. Solar Energy Laboratory,
University of Wisconsin, Engineering Experimentation Station, Madison, Wisconsin, September 1976.