Case Studies in Thermal Engineering 17 (2020) 100569

Contents lists available at ScienceDirect

Case Studies in Thermal Engineering

journal homepage: http://www.elsevier.com/locate/csite

Energy savings approaches for high-tech manufacturing factories

Shih-Cheng Hu a, Tee Lin a, Shao-Huan Huang a, Ben-Ran Fu b, *, Ming-Hsuan Hu c
a
Department of Energy and Refrigerating Air-conditioning Engineering, National Taipei University of Technology, Taipei, 106, Taiwan
b
Department of Mechanical Engineering, Ming Chi University of Technology, New Taipei City, 243, Taiwan
c
Department of Energy Technology, Royal Institute of Technology (KTH), Stockholm, Sweden

A R T I C L E I N F O A B S T R A C T

Keywords: This study integrates a fab energy simulation (FES) tool and energy conversion factors to analyze
Fab energy simulation energy consumption and identify energy savings opportunities of high-tech manufacturing fac­
Energy conversion factors tories. The data used is the 169,124 MW h annual energy consumption of a Taiwanese semi­
Energy consumption
conductor manufacturing fab. We proposed a compressed/clean dry air (CDA) system as a two-
Energy savings
Semiconductor manufacturing
pressure system using a heated-type dryer. The proposed method achieved 3050 MW h as the
highest energy savings in the CDA system compared to the original fab data. The results also
indicate that lowering the make-up air unit (MAU) operating temperature caused significant
energy savings for the high-temperature water chiller system. When the outlet air temperature of
the MAU system was 14 � C, the maximum energy savings were 3532 MW h. In addition, the
process cooling water (PCW) system could potentially save energy by reducing the pumping head
of water in its open system. Therefore, we suggested a closed PCW system that reached a
maximum energy saving of 1541 MW h.

1. Introduction

The heating, ventilating, and air conditioning (HVAC) systems used in high-tech fabs generally prioritize clean water and air supply
of buildings, resulting in energy consumption characteristics different from those in commercial buildings. Fabs consume more energy
due to rigorous requirements for temperature, relative humidity, and particle contamination. Consequently, researchers have studied
opportunities to save energy in the HVAC system of high-tech industries. Yu et al. [1] proposed the variable control methodology for
air re-circulation units to enhance the efficiency of HVAC system energy in fabs. The results indicated that when their approach in the
studied fab reduced temperature and humidity variations—indicating stable system operation—fan, heater, and water pump energy
consumption fell 31.0%, 70.7%, and 30.6%, respectively. Su and Yu [2] developed an approach for evaluating the differential pressure
set point for chilled water pumps in the HVAC system of a cleanroom and found that when the pressure set point obtained from their
approach was employed, the energy savings were approximately 53%.
Chang et al. [3] explored energy-saving approaches for TFT-LCD factories in Taiwan and used a fab energy system (FES) developed
by Hu et al. [4]. Their results demonstrated that adjusting the set points of the dry bulb temperature and relative humidity in the
cleanroom had the highest influence on energy consumption. The highest energy saving was approximately 1% of the baseline in the
studied fab. Shan and Wang [5] investigated the design and control of cleanroom air-conditioning systems. Their results demonstrated
that air humidity, temperature, and pressure in cleanrooms could be controlled using the partially decoupled option they proposed.

* Corresponding author.
E-mail addresses: brfu@mx.nthu.edu.tw, brfu@mail.mcut.edu.tw (B.-R. Fu).

https://doi.org/10.1016/j.csite.2019.100569
Received 22 October 2019; Received in revised form 20 November 2019; Accepted 22 November 2019
Available online 23 November 2019
2214-157X/© 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
S.-C. Hu et al. Case Studies in Thermal Engineering 17 (2020) 100569

Compared to their reference method, the interactive approach, their method reduced electricity and gas consumption for cooling and
heating/reheating by 69.6% and 87.8%, respectively. In addition, no hardware modification was required for this in the system
retrofit. Tsao et al. [6] studied energy saving from a make-up air unit (MAU) system used in a cleanroom as it can account for
approximately 50% of the system energy consumption of the chiller [7]. Of the different arrangements studied, the dry cooling coil
(DCC) system integrated with a water return approach obtained the highest energy efficiency.
Compressed/clean dry air (CDA) is employed in fabs to drive pneumatic components of manufacturing process equipment, control
valve functioning, or clean materials/equipment. To smooth the manufacturing processes, CDA system energy efficiency is neglected
during operation, resulting in unnecessary wasted energy. Saidur et al. [8] reported that the energy use efficiency of a clean/com­
pressed air system may be only 10–20% and that energy loss was mainly from heat dissipation and air leakage during the process. Lin
[9] discovered while the dew point temperature of a CDA system increased from 70 � C to 20 � C (heatless dryer), the unit energy
consumption for the CDA system decreased from 147 W/m3 to 132 W/m3, a 7.3% saving. The daily energy consumption of the CDA
with an additional thermal dryer was reduced by 20.8%.
Semiconductor Equipment and Materials International (SEMI) published the semiconductor facility systems guidelines (SEMI S23-
0813) for energy, electricity, and production conservation [10]. SEMI S23-0813 provides the energy conversion factors (ECFs, energy
consumption per unit flow rate) of important utilities, summarized in Table 1. The ECFs estimate the energy consumption of utilities
and is used to discover energy savings at high-tech fabs. Hu et al. [11] applied the SEMI ECFs to semiconductor and LCD manufacturing
plants and developed a new ECF calculator with corresponding mathematical models for each subsystem or utility. Their calculated
ECF values were close to SEMI’s in the same operating conditions. Their ECF calculator can determine energy savings and clarify the
detailed contribution of each variable to the energy consumption of each system in a fab.
Currently, high-tech product development focuses on multiple functionality and miniaturization. Electronic component functions
are becoming increasingly complex, wafer-process-line size is shrinking, and panel sizes are increasing. Therefore, precision in factory
systems is important. Cleanliness levels and product manufacturing yield are related to the cleanroom, and subsidiary facilities in
laboratories and production regions are also important. Semiconductor and panel manufacturing factories have many similarities in
production processes and facilities. The facilities are divided into cleanroom, air-conditioning, power/instrumentation systems,
process cooling water (PCW), CDA, ultra-pure water (UPW), exhaust, and gas supply systems. Manufacturers aim for every link to
perform efficiently and seek energy savings opportunities to reduce production costs and maintain competitiveness. Considering these
issues, the objective of this study is complementary to our previous work [4,11] and aims to integrate the FES tool and ECFs to analyze
the energy use of high-tech fabs. We expected to discover energy savings opportunities from the results.

2. Methodology

2.1. FES tool and ECF calculator

Energy consumption in a fab is divided into several parts, including the HVAC system, the exhaust system, PCW, UPW, CDA, ni­
trogen, vacuum, fans, pumps, process tools, and lighting system. These facilities and materials are considered in the FES described in
Ref. [4]. In Ref. [11], an ECF calculator was developed, based on SEMI S23-0813, to understand the energy consumption in fabs with

Table 1
Energy conversion factors (ECFs) suggested in SEMI S23-0813 [10].
Facility (sub-system) or material Unit ECF Condition or remark
3
a. Exhaust kWh/m 0.0037 Exhaust pressure ¼ 2 kPa.
b. Vacuum kWh/m3 0.060 Vacuum pressure ¼ 5.88 kPa.
c. Clean dry air (CDA) kWh/m3 0.147*
d. High pressure CDA (for 827–1034 kWh/m3 0.175*
kPa, gauge)
e. Water cooled by refrigeration (for kWh/m3 1.56 Supply pressure ¼ 490 kPa. For other ΔT, ECF ¼ 0.258 � ΔT þ 0.273.
ΔT ¼ 5 � C)
f. Water cooled by cooling-tower kWh/m3 0.260 Supply pressure ¼ 490 kPa.
g. UPWa or DIWa (for temperature < 25 kWh/m3 9.0 Supply pressure ¼ 196 kPa.
�
C)
h. Hot UPW or DIW (for temperature > kWh/m3 92.2 No suggested ECF for the temperature of 25–85 � C.
85 � C)
i. Heat load
Heat removal via air kWh/m3 3.24 � This conversion factor accounts for the energy that may be used to operate the cleanroom
�
C 10 4 air conditioning.
Heat removal via water kWh/m3 1.16
�
C
Cooling load kWh/m3 0.287
�
C
j. N2 kWh/m3 0.250* Supply pressure ¼ 793 kPa.
k. Electricity kWh/ 1.0 This ECF can be further modified by accounting for the efficiency of the generating
kWh electricity through different ways.
a
UPW: ultra-pure water; DIW: de-ionized water. * On the basis of the equivalent volume at the pressure of 101 kPa and temperature of 20 � C.

2
S.-C. Hu et al. Case Studies in Thermal Engineering 17 (2020) 100569

different operating conditions and/or scale. The detailed description and mathematical models for the systems are described in Refs.
[4,11]. In the FES tool, the MAU system can be rearranged according to the needs of a fab, and three control methods can be employed
for each component (i.e., dry-point temperature control, dew-point temperature control, and enthalpy control). For practical appli­
cations, the coils integrated with pre-cooling, pre-heating, and re-heating functions are used to control the dry-point temperature; the
second cooling coil, with washer-type humidification and fans, is used to control the dew-point temperature. Each control variable can
be changed at any time in the FES. Therefore, the FES can accurately model the MAU energy consumption system. The ECF may be
required for the FES when exact operating data are absent. In the present study, the ECFs for the exhaust system, PCW, UPW, CDA,
nitrogen, and vacuum were estimated using the measured data of the studied fab or the ECF calculator developed by Ref. [11].
Therefore, we obtained detailed energy consumption for a fab by integrating the FES tool and the ECF calculator with measured data.

2.2. Details of the fab

We used data from a semiconductor manufacturing fab in Hsinchu, Taiwan, to demonstrate the integration and application of the
FES tool and the ECF calculator. Eight MAU systems were used, and detailed data is summarized in Table 2. The temperature and
relative humidity in the cleanroom of the fab were 23 � C and 45%, respectively. The lighting intensity was 0.0119 kW/m2 with a 60%
loading factor, and the cooling load of each worker estimated as 0.16 kW. Fig. 1 details the annual energy consumption of fab
components as measured and collected for the present study, along with component energy consumption proportion markings. The
annual energy consumption was 169,124 MW h and the annual electricity cost is approximately 13,870,000 USD (Taiwan electricity
use price: 0.082 USD/kWh). Based on the measured data, the process equipment was approximately 39.0% of the total energy con­
sumption, proportional the report by Hu and Chuah [7]. In addition to the process equipment, the highest energy consumers in the fab
were the CDA (22.1%), water chiller (13.4% þ 7.2% ¼ 20.6%), PCW (5.8%), de-ionized water (DIW) (3.2%), and exhaust systems
(2.0%).

2.3. Approaches to investigate energy savings

Four approaches were employed to find potential energy savings. They corresponded to the highest energy consumers in the fab
(excluding the DIW system due to the modification difficulty and working conditions limitations). In approach A, we divided the CDA
single supply-pressure system into a dual supply-pressure system and replaced the non-heated adsorption dryer with a heated
adsorption dryer with a blower. In approach B, we lowered the outlet air temperature of the MAU system to explore the effect on the
water chiller energy consumption. In approach C, we explored the effects of temperature differences and pumping head of the PCW
system on energy consumption. In approach D, we reduced the supply flow rate of the MAU system. The four approaches, detailed in
Table 3, were chosen because they significantly influenced the energy consumption of the fab and required minor modifications on the
original system, easily implemented in a fab. Approaches A-1, B-1, C-1, and D-1 in Table 3 performed under the same operating
condition of the fab and were used as reference cases.

3. Results and discussion

3.1. ECFs for the fab

Table 4 shows the calculated ECFs for the main components of the fab (based on the measured data) corresponding to its working
flow rate. The ECF for the CDA system was 0.154 kW h/m3, similar to the value in SEMI S23-0813 (4.8% deviation). However, the ECFs
for the PCW, DIW, exhaust, and UPW systems were lower than the SEMI ECFs. For example, the PCW ECF of the fab and the SEMI
suggested value were 0.994 and 1.56 kW h/m3, respectively. The fab PCW’s ECF was 36.3% lower than the SEMI ECF. This result
indicates that the SEMI ECFs may not be suitable for a fab with different operating conditions. For this reason, Hu et al. [11] developed
a new ECF calculator.

Table 2
The information of the MAU system for the studied fab.
Parameter or condition Information

Outer air condition From the weather report in Hsinchu, Taiwan

Air temperature at the outlet of pre-heating coil 14 � C
Air temperature at the outlet of pre-cooling coil 20 � C
Humidification Washer-type
Dew-point temperature for the cooling coil 10.5 � C
Relative humidity for the cooling coil 100%
Air temperature at the outlet of re-heating coil 19 � C
Operating flow rate of fan 150,000 m3/h
Static pressure of fan 1180 Pa
Motor efficiency 80%
Fan efficiency 83%

3
S.-C. Hu et al. Case Studies in Thermal Engineering 17 (2020) 100569

Fig. 1. Annual energy consumption of components in the studied fab.

Table 3
Conditions of the studied approaches.
Approach Original condition Modification

A-1 non-heated adsorption dryer and single-pressure supply ¼ 7.5 kg/

cm2
A-2 heated adsorption dryer with blower
A-3 dual-pressure supply ¼ 5.5 and 7.5 kg/cm2
A-4 heated adsorption dryer with blower and dual-pressure supply ¼ 5.5 and 7.5
kg/cm2
B-1 outlet air temperature of the MAU system ¼ 19 � C
B-2 outlet air temperature of the MAU system ¼ 18 �
C
B-3 outlet air temperature of the MAU system ¼ 17 �
C
B-4 outlet air temperature of the MAU system ¼ 16 �
C
B-5 outlet air temperature of the MAU system ¼ 15 �
C
B-6 outlet air temperature of the MAU system ¼ 14 �
C
C-1 temperature difference ¼ 3.2 � C and pumping head ¼ 56 m
C-2 temperature difference ¼ 4 � C
C-3 pumping head ¼ 34 m
C-4 temperature difference ¼ 4 � C and pumping head ¼ 34 m
D-1 supply flow rate of the MAU system ¼ 213,304 m3/h
D-2 reduced by 5%, i.e., flow rate ¼ 202,639 m3/h
D-3 reduced by 10%, i.e., flow rate ¼ 191,974 m3/h

Table 4
ECFs for main components of the studied fab.
Component Flow rate (m3/h) Calculated ECF (kWh/m3) SEMI ECF (kWh/m3) Deviation to SEMI ECF

CDA 27,593 0.154 0.147 4.8%

PCW 1122 0.994 1.56 36.3%
DIW 103 5.99 9.0 33.4%
Exhaust 213,304 0.00142 0.0037 61.6%
UPW 58 4.34 9.0 51.8%

3.2. The CDA system

The CDA system of the fab included eight air compressors, three air storage tanks, eight dryers, and three filters. The effect of
different dryer types on the energy consumption of the CDA system was investigated. The non-heated adsorption dryer is not heated
during the regeneration process, whereas the heated adsorption dryer is heated during this process. A non-heated adsorption dryer was
originally used in the studied fab. Generally, a non-heated adsorption dryer provides approximately 15%–20% of the CDA into the CDA
regeneration tank system for drying purposes (also called purge loss), indicating no additional heating and blowing equipment is
needed. However, the system operation costs are high due to the CDA consumed during its generation period. A heated adsorption
dryer requires a blower and heater to heat the air during the regeneration process of desiccant. The air is heated to 180 � C to desorb the
humidity in the desiccant. A heated adsorption dryer also consumes approximately 5% of the generated CDA for cooling the desiccant
with a longer cycle time than a non-heated dryer. Table 5 lists the design parameters imported into the ECF calculator to calculate the
corresponding ECFs for these two types of dryers; the ECFs were used as inputs for the FES. The calculated ECFs for the non-heated and

4
S.-C. Hu et al. Case Studies in Thermal Engineering 17 (2020) 100569

heated dryers were 0.0139 and 0.0111 kW h/m3, respectively, under design conditions. These results indicate a 20% energy savings on
the heated dryer with a blower compared to a non-heated dryer.
The on-site CDA survey used in the fab indicated that nine process tools consumed CDA with different operating pressures,
summarized in Fig. 2. However, the supply pressure of the CDA system of the fab was only at one level (7.5 kg/cm2) with total flow rate
requirement of 27,593 m3/h. Consequently, the CDA system wasted energy due to the mismatch between the operating pressure and
the pressure required for different process tools. Dividing the CDA supply pressure into different levels may reduce this energy loss.
Based on the survey, we classify the CDA supply pressure as 5.5 kg/cm2 (used for operating pressure lower than or equal to 5.5 kg/cm2)
and 7.5 kg/cm2 (used for operating pressure higher than 5.5 kg/cm2). Thus, the corresponding flow rate requirements are 7004 and
20,589 m3/h. For the CDA system with a non-heated dryer, the calculated ECFs for the 5.5 and 7.5 kg/cm2 supply pressure levels were
0.129 and 0.154 kW h/m3, respectively. The ECFs were 0.121 and 0.143 kW h/m3 for the CDA system with a heated dryer. Table 6
shows the calculated ECFs details for the CDA system with two supply-pressure levels and a heated dryer (i.e., approach A-4).
Fig. 3 summarizes annual energy savings for the CDA system in the A-2 to A-4 approaches compared to the original data (approach
A-1). The highest energy savings of the CDA system (approach A-4) was 3050 MW h (the CDA system energy consumption reduced by
8.17%). The annual energy savings for approaches A-2 and A-3 were 1696 and 1538 MW h (reductions of 4.54% and 4.12%),
respectively. The overall energy savings for the fab for the approaches A-2, A-3, and A-4 were 1.00%, 0.91%, and 1.81%, respectively.
Based on our survey for the commercial equipment, implementing approach A-4 in the fab costs approximately 430,000 USD for eight
heated-dryers and the annual cost reduction from electricity savings is approximately 250,000 USD. The estimated payback of the
dryer investment was 1.72 years, considered acceptable for most high-tech manufacturing businesses.

3.3. The MAU system

The outlet air temperature in the MAU system was 19 � C that could be further reduced. The air psychrometric chart suggested that
outlet temperatures less than 13 � C should be avoided because low temperatures may lead to frost or mildew on the HEPA (high
efficiency particulate air) filter in the MAU system (causing malfunction). In addition, the relative humidity for a HEPA filter should be
lower than 80% during long-term operation. Therefore, we studied the effects of outlet air temperature of the MAU system on the water
chiller energy consumption by reducing the outlet air temperature from 19 � C to 14 � C and 1 � C interval (corresponding to approaches
B-1 to B-6). In general, lowering the outlet air temperature reduces the load requirements of the reheating coil and the DCC. However,
pressure loss through the HEPA filter may also increase as air temperature decreases.
The fab source of heat was from a high-temperature water chiller with heat recovery function. Thus, the temperature change in the
MAU system affects mainly the performance of the high-temperature water chiller. Fig. 4 shows the high-temperature water chiller
system annual energy savings with changes to the MAU system outlet air temperature. The energy consumption decreased linearly
from 12,134 MW h to 8602 MW h as the MAU system outlet air temperature decreased from 19 � C to 14 � C. Lowering the temperature
by 1 � C corresponded to a 5.82% reduction in energy consumption in the high-temperature water chiller system. For approach B-6, the
14 � C outlet air temperature case, energy savings reached a maximum of 3532 MW h, approximately 29.1% of the original data
(approach B-1). It also indicated 2.09% of energy savings based on overall energy consumption in the fab. In addition, the FES results
can show the detailed energy or heat flow in the fab. For example, sensible heat removal in the cleanroom of the fab for approaches B-1
to B-6 increased linearly from 49.1 MW h to 68.1 MW h. The corresponding load of the pre-heating coil in the MAU system decreased
form 32.2 MW h to 13.2 MW h. These results demonstrate significant energy savings from lowering the operating temperature. Most
importantly, the implement of approach B does not require additional equipment. The annual cost reduction from electricity savings is
approximately 290,000 USD when implementing approach B-6.

3.4. The PCW system

In the fab, the heat generated by process tools was removed with the PCW system requiring temperatures between 1722 � C. In
general, the PCW is provided (or available via heat exchange) by a water chiller (refrigeration system) or cooling tower, depending on

Table 5
The design parameters of two studied types of the dryer.
Parameter Non-heated type Heated type with blower

Flow rate (m3/min) 80 80

Ambient temperature (� C) 35 35
Inlet temperature (� C) 25 25
Inlet pressure (kg/cm2) 7.5 7.5
Dew-point temperature (� C) 70 70
Adsorption efficiency 10% 10%
Water amount (kg/m3) 0.0019 0.0025
Purge loss 15% 5%
Switch period 15 min 8h
Heating time – 3h
Blower power (kW) – 11.2 (15 HP)
Heating power (kW) – 72

5
S.-C. Hu et al. Case Studies in Thermal Engineering 17 (2020) 100569

Fig. 2. The CDA requirement with different operating pressures for nine process tools in the studied fab.

Table 6
Calculated ECFs for the CDA system with two supply-pressure levels and a heated dryer (approach A-4).
Utility Unit 5.5 kg/cm2 7.5 kg/cm2

A. Air compressor
A1. Suction pressure kPa 101 101
A2. Discharge pressure kPa 640 836
A3. Polytropic index n during the process – 1.4 1.4
A4. Theoretical compressor input work kWh/m3 0.068 0.081
A5. Adiabatic efficiency – 85% 85%
A6. System efficiency – 90% 90%
A7. Energy consumption of the compressor kWh/m3 0.089 0.106
B. Dryer
B1. Purge loss – 5% 5%
B2. Energy consumption of the dryer kWh/m3 0.00648 0.00648
C. Cooling equipment
C1. Heat dissipation for the compressor kWh/m3 0.094 0.112
C2. Heat dissipation for the dryer kWh/m3 0.00648 0.00648
C3. Energy needed for the unit heat dissipation kWh/kWh 0.195 0.195
C4. Energy consumption for the cooling system kWh/m3 0.020 0.023

ECF ¼ (A7 þ B2 þ C4)/(1 – B1) kWh/m3 0.121 0.143

Fig. 3. Annual energy savings for the CDA system under approaches A-2 to A-4.

the temperature requirement. As listed in Table 1, the SEMI ECFs for the PCW system supplied by chiller and cooling tower are 1.56
and 0.260 kW h/m3, respectively. The studied PCW system in the fab had an open-type water reservoir cooled by a water chiller.
Although the PCW ECF for the fab was 0.994 kW h/m3, significantly lower than the SEMI ECF, we determined the PCW system may
have potential energy savings due to the water pumping loss in its open system. Another issue for open-type PCW systems is supply
water shortage in pipelines during a power outage or system shutdown (process tools damage). Therefore, we studied modifications for
a closed system in which the pumping head was decreased from 54 m to 36 m. We found that energy consumption for pumping power
decreased. In addition, we investigated the PCW system temperature difference effects (3.2 � C or 4 � C) on energy consumption. The

6
S.-C. Hu et al. Case Studies in Thermal Engineering 17 (2020) 100569

Fig. 4. Annual energy savings for the high-temperature water chiller system under approaches B-2 to B-6.

results for approaches C-1 to C-4 are summarized in Fig. 5. The energy savings for approaches C-2, C-3, and C-4 were 0.43%, 0.62%,
and 0.91%, respectively, of the overall energy consumption in the fab, corresponding to ECFs of 1.175, 0.888, and 1.071 kW h/m3.
Although energy consumption in approaches C-2 and C-4 was reduced, the PCW ECFs were higher than the original ECF of the fab
because the temperature difference of the PCW for these two cases increased from 3.2 � C to 4 � C, and the requirement of its flow rate
reduced by 20% from the same requirement of the heat-removal load. The maximum savings was 1541 MW h (approach C-4), and
15.7% energy savings in the PCW system. The maximum energy savings reduces the annual electricity use cost by 126,000 USD if
neglecting the modification cost of the piping system.

3.5. Exhaust system

We explored the effect of flow rate of the exhaust system on energy consumption, approaches D-1 (original case), D-2 (reduced by
5%), and D-3 (reduced by 10%). In this study, the exhaust system refers to a general type (i.e., no acidic, alkaline, toxic, and volatile
exhausts). In the studied fab, the exhaust was removed by fans in the cleanroom. The decreased flow rate lowered exhaust fan energy
consumption. The approach D had significantly different results from approaches A, B, and C. Not only the exhaust fans, and also other
systems or heat-load requirements (e.g., water chillers, fans in MAU system, sensible heat removal in the cleanroom, cooling, pre-
heating, and re-heating loads in the MAU system) were influenced by changes in the flow rate. For example, as the exhaust system
flow rate decreased by 5% or 10%, the exhaust fan energy consumption decreased by 133 MW h (3.9%) or 266 MW h (7.8%), and the
high-temperature water chiller energy consumption decreased by 19 MW h (0.16%) or 40 MW h (0.33%). The corresponding load of
the re-heating coil from the MAU system decreased by 272 MW h (0.85%) or 565 MW h (1.76%). Table 7 summarizes the results for this
approach and shows that overall energy savings for approaches D-2 and D-3 were 0.16% and 0.32%, respectively. Additionally, no
additional equipment is required to employ approach D. The annual cost reduction from electricity savings is approximately 44,800
USD when implementing approach D-3. Furthermore, approach D is not compatible with other studied approaches because of the
different baseline applied.

4. Conclusions

This study complements our previous studies [4,11] and demonstrates the integration and application of the FES tool and the ECF
calculator. The energy performance of the utility system of a fab can be realized and the specific influences on each system can be
analyzed. The following may be concluded based on the results of this study:

(1) The ECF of the CDA system was 0.154 kW h/m3, approximately equivalent to the suggested value in SEMI S23-0813; the ECFs
for the PCW, DIW, and the exhaust systems were significantly lower than the SEMI ECFs.
(2) The CDA system in the fab was classified as a two pressure levels system using a heated-type dryer. Compared to the original
data, the highest energy savings for the CDA system (i.e., approach A-4) was 3050 MW h (i.e., the energy consumption in the
CDA system reduced by 8.17%); this energy savings constituted 1.81% of the overall energy consumption of the fab.
(3) Lowering the MAU operating temperature caused significant energy savings for the high-temperature water chiller system.
When the outlet air temperature was 14 � C (approach B-6), the energy savings reached a maximum value of 3532 MW h,
corresponding to approximately 29.1% of the original data, and it constituted an energy savings of 2.09% of the overall energy
consumption in the fab.
(4) Although the PCW ECF for the studied fab was 0.994 kW h/m3, significantly lower than the SEMI ECF, we expected energy
savings in the PCW system due to water pumping loss of the open system. Therefore, we suggested a closed system. The
maximum energy saving was 1541 MW h (approach C-4), a PCW system energy savings of 15.7% (0.91% of overall fab energy
consumption).
(5) The results also demonstrated that not only exhaust fans, and other systems or heat-load requirements, were influenced during
flow rate changes to the exhaust system. For example, as the flow rate of the exhaust system decreased by 10% (approach D-3),

7
S.-C. Hu et al. Case Studies in Thermal Engineering 17 (2020) 100569

Fig. 5. Annual energy savings for the PCW system under approaches C-2 to C-4.

Table 7
Energy consumption or affected parameter under approaches D-1 to D-3.
Energy consumption or affected parameter Approach D-1 Approach D-2 Approach D-3

Exhaust fans (MWh) 3382 3249 3116

High-temperature water chiller (MWh) 12,134 12,115 12,094
Low-temperature water chiller (MWh) 22,637 22,521 22,397
MAU’s fans (MWh) 7477 7413 7345
Sensible heat removal in the clean room (MWh) 49,142 48,952 48,756
MAU cooling load (MWh) 106,689 105,786 104,816
MAU pre-heating load (MWh) 1193 1183 1172
MAU re-heating load (MWh) 32,157 31,885 31,592
Total energy consumption in the fab (MWh) 169,124 168,856 168,578
Energy savings for the fab – 0.16% 0.32%

the exhaust fans energy consumption decreased by 266 MW h (7.8%); the high-temperature water chiller energy consumption
decreased by 40 MW h (0.33%); in addition, the corresponding load of the re-heating coil in the MAU system decreased by 565
MW h (1.76%). However, overall energy savings of approach D-3 was only 0.32%.

Taiwan maintains the largest share of semiconductor wafer fab capacity in the world since 2016, as 21.3% of worldwide in 2016
[12]. Therefore, a validation by a fab data in Taiwan has symbolic value. The replication of this work is possible for other fabs in the
world, and the present methodology can be generalized and applied to other fabs. The differences among different fabs might be
operational parameters, component efficiencies, and outside environment conditions. We can modify those differences in the present
analysis process to obtain the corresponding results.

Author contribution statement

Shih-Cheng Hu: Conceptualization, Investigation, Supervision. Tee Lin: Conceptualization, Methodology, Investigation. Shao-Huan
Huang: Conceptualization, Methodology, Investigation. Ben-Ran Fu: Investigation, Writing- Original draft preparation, Reviewing,
and Editing. Ming-Hsuan Hu: Conceptualization, Methodology.

Declaration of competing interest

We confirm that there are no known conflicts of interest associated with this publication and there has been no influence for this
work that could have influenced its outcome.

Acknowledgments

This work was supported by the Ministry of Science and Technology, Taiwan (grant number: MOST 106-2221-E-027-111).

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.csite.2019.100569.

8
S.-C. Hu et al. Case Studies in Thermal Engineering 17 (2020) 100569

References

[1] K.T. Yu, C.L. Su, J.L. Kuo, Variable recycled air controls of HVAC systems for energy savings in high-tech industries, IEEE Int. Conf. Ind. Technol. (ICIT) (2016)
215–220. Taipei, Taiwan, March 14–17, 2016.
[2] C.L. Su, K.T. Yu, Evaluation of differential pressure setpoint of chilled water pumps in cleanroom HVAC systems for energy savings in high-tech industries, in:
IEEE/IAS 48th Industrial & Commercial Power Systems Technical Conference (I&CPS), Louisville, KY, USA, May 20–24, 2012, 2012.
[3] C.K. Chang, T. Lin, S.C. Hu, B.R. Fu, J.S. Hsu, Various energy-saving approaches to a TFT-LCD panel fab, Sustainability 8 (2016) 907.
[4] S.C. Hu, T. Lin, B.R. Fu, C.K. Chang, I.Y. Cheng, Analysis of energy efficiency improvement of high-tech fabrication plants, Int. J. Low Carbon Technol. (2019),
https://doi.org/10.1093/ijlct/ctz041.
[5] K. Shan, S. Wang, Energy efficient design and control of cleanroom environment control systems in subtropical regions – a comparative analysis and on-site
validation, Appl. Energy 204 (2017) 582–595.
[6] J.M. Tsao, S.C. Hu, T. Xu, D.Y.L. Chan, Capturing energy-saving opportunities in make-up air systems for cleanrooms of high-technology fabrication plant in
subtropical climate, Energy Build. 42 (2010) 2005–2013.
[7] S.C. Hu, Y.K. Chuah, Power consumption of semiconductor fabs in Taiwan, Energy 28 (2003) 895–907.
[8] R. Saidur, N.A. Rahim, M. Hasanuzzaman, A review on compressed-air energy use and energy savings, Renew. Sustain. Energy Rev. 14 (2010) 1135–1153.
[9] Y.C. Lin, Energy-saving Study of the Air Condition and Air Compressor Equipment for TFT-LCD Factory, Master Thesis, Department of Energy and Refrigerating
Air-conditioning Engineering, National Taipei University of Technology, Taipei, Taiwan, 2013 (in Chinese).
[10] SEMI S23-0813, Guide for conservation of energy, utilities and materials used by semiconductor manufacturing equipment, Semicond. Equip. Mater. Int.
(August 2013).
[11] S.C. Hu, Y.W. Tsai, B.R. Fu, C.K. Chang, Assessment of the SEMI energy conversion factor and its application for semiconductor and LCD fabs, Appl. Therm. Eng.
121 (2017) 39–47.
[12] Taiwan Maintains Largest Share of Semiconductor Wafer Fab Capacity, announced by AnySilicon on February 25, 2017. Available from, https://anysilicon.
com/taiwan-maintains-largest-share-semiconductor-wafer-fab-capacity/. (Accessed 5 April 2019).