Professional Documents
Culture Documents
Energy Volume 107 Issue 2016 (Doi 10.1016/j.energy.2016.03.139) Kim, Wonuk Jeon, Seung Won Kim, Yongchan - Model-Based Multi-Objective Optimal Control of A VRF (Variable Refrigerant Flow) Combine
Energy Volume 107 Issue 2016 (Doi 10.1016/j.energy.2016.03.139) Kim, Wonuk Jeon, Seung Won Kim, Yongchan - Model-Based Multi-Objective Optimal Control of A VRF (Variable Refrigerant Flow) Combine
Energy
journal homepage: www.elsevier.com/locate/energy
Department of Mechanical Engineering, Korea University, Anam-Dong, Sungbuk-Gu, Seoul, 136-713, Republic of Korea
Korea Institute of Civil Engineering and Building Technology, 64, 182 Beon-Gil, Mado-Ro, Mado-Myeon, Hwaseong-Si, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 19 November 2015
Received in revised form
10 March 2016
Accepted 30 March 2016
A VRF (variable refrigerant ow) combined system adopting a DOAS (dedicated outdoor air system) has
been proposed to reduce the total energy consumption while satisfying IAQ (indoor air quality) and THC
(thermal and humidity comfort) with minimum outdoor air. The objective of this study is to develop a
model-based multi-objective optimal control strategy for the VRF combined system with multi-zone in
order to optimize the multi-objective functions of the THC, IAQ, and total energy consumption. The
performance of the VRF combined system was evaluated using the EnergyPlus model. The VRF combined
system was optimized by GA (genetic algorithm) and RSM (response surface methodology) with the
multi-objective functions of the THC, IAQ, and total energy consumption. The proposed multi-objective
optimal control strategies (A and B) were compared with the TS (time schedule) strategy and the DCVH
(demand controlled ventilation with humidifying). Optimal control strategy B reduced the total energy
consumption by 20.4% and increased the ratio of the hours satisfying the extended comfort zone by 19.1%
compared to the DCVH strategy.
2016 Elsevier Ltd. All rights reserved.
Keywords:
Variable refrigerant ow
Dedicated outdoor air system
Model-based control
Multi-objective optimization
Multi-zone ventilation
1. Introduction
Nowadays, building energy management is strongly inuenced
by IAQ (indoor air quality) and thermal comfort in working spaces.
While ventilation with fresh OA (outdoor air) is essential for maintaining acceptable IAQ, it increases the thermal load of the working
space. New ventilation systems using novel control methods such as
DOAS (dedicated outdoor air system) and DCV (demand controlled
ventilation) have been proposed to reduce the total energy consumption while satisfying IAQ with minimum OA. The conventional
centralized all-air system consumes substantial energy to ventilate
the air in a building for thermal comfort and IAQ, but the DOAS can
reduce energy consumption by ventilating air only for IAQ [1]. In
addition, the DOAS can allow easy control of IAQ by separating
ventilation air from heating and cooling air [2]. The DOAS is often
operated using DCV (demand controlled ventilation). The DCV controls ventilation airow to satisfy the balance between IAQ and energy saving [3]. Therefore, the DOAS has been considered as an
alternative to the centralized all-air system in commercial buildings.
Nomenclature
C
CO2 concentration, ppm
COP
coefcient of performance
CR
combination ratio
Cv(RMSE)coefcient of variation of the root mean squared error
DCV
demand controlled ventilation
DCVH
DCV with a humidier
DOAS
dedicated outdoor air system
E
energy consumption, MJ
EA
exhaust air
EEV
electronic expansion valve
EIR
energy input ratio
GA
genetic algorithm
HVAC
heating, ventilating, and air-conditioning
IAQ
indoor air quality
IDU
indoor unit
L
length, m
m
mass ow rate, kg s1
MBE
mean bias error
O
occupancy
OA
outdoor air
Obj
objective function
ODU
outdoor unit
PLR
part-load ratio
PMV
predicted mean vote
R2
coefcient of determination
RA
return air
RH
relative humidity, %
RSM
SA
T
THC
TS
VAV
VRF
wf
197
Subscripts
DOAS
dedicated outdoor air system
E
total energy consumption
high
high humidity zone
humidier humidier
i
ith air conditioned zone
IAQ
indoor air quality
IDB
indoor air dry-bulb
in
indoor
max
maximum
min
minimum
out
outdoor
OWB
outdoor air wet-bulb
p
pipe
SA
supply air
sp
set point
sup
superheat
THC
thermal and humidity comfort
VRF
variable refrigerant ow
VRF heat pump, a DOAS with a plate type total heat exchanger, and
an electrically heated steam humidier. The VRF system was
designed to cover the sensible load in the conditioned multi-zone.
The VRF system consisted of a single ODU (outdoor unit) and multiIDUs for multi-zone. The VRF system can handle various heating/
cooling loads in multi-zone by controlling a variable speed
compressor and EEV (electronic expansion valve) according to
refrigerant superheat, indoor temperature, and occupancy. The
heating and cooling capacities of the VRF system were enhanced by
increasing the compressor speed, and the mass ow rate through
the VRF system was controlled by the EEV [17]. The DOAS with
energy recovery was used as a separate ventilation unit to change
indoor air with fresh OA. The DOAS reduced the additional heating/
cooling loads resulting from in-taking OA by the energy recovery
between EA (exhaust air) and OA [1]. The DOAS with the humidier
was used to cover the latent load and IAQ in the conditioned multizone. An electrically heated steam humidier was installed between the zone splitter and the outlet of the DOAS system.
The VRF combined system was installed in a building with 7 aboveground oors and 1 underground oor, which is located in Seoul,
Korea. Table 1 shows the details of the building. As shown in Fig. 2, the
oor plan consists of six zones, a hallway, an elevator, two stairs, and a
toilet. Only one oor with the six zones and a hallway was considered
in the simulation. The thermal conductivity of the building envelope
and the internal heat gain from the light and equipment are given in
Table 1. Table 2 shows the specications of the VRF system and DOAS.
The heating capacity and COP (coefcient of performance) of the VRF
system were 65.2 kW and 4.2, respectively, at the rated conditions. The
heating sensible effectiveness and heating latent effectiveness of the
DOAS were 76% and 54%, respectively, for the rated 100% ow condition and 79% and 56%, respectively, for the rated 75% ow condition.
The effectiveness was determined by linear interpolation or
198
Humidity
controller
DOAS
EA
Supply fan
SA
RA
OA flow
rate
controller
EEV
controller
Tsp reset
controller
Tsup Tin RH O
Tout
Refrigerant
flow
controller
IDU 1
Zone 1
Tsup Tin RH O
VRF
Outdoor unit
Tsup Tin RH O
IDU i
Constraints
Damper
Damper
Zone 2
...
IDU 2
Genetic
algorithm
Iterative calls
during
optimizing
Data from
predictor
...
OA
Multi-objective
function
estimator
User-define
weight factors
Damper
Zone i
Model-based predictor
(EnergyPlus)
Multi-objective optimizer
(MATLAB)
Table 1
Brief description of the case building.
Section
Details
Location
Building
Total oor area
Plan and height
Operating/ofce hours
HVAC design parameter
Building envelope
External wall
Roof
Floor
Window
Note: Based on the Korean building regulation, the minimum ventilation air ow rate requirement for the ofce building is 34.2 m3 h1 per person. In the ventilation air ow
rate, the exhaust air from the kitchen and bathrooms can be included. In addition, the minimum occupant load requirement is 9.3 m2 per person.
extrapolation from the values for the 100% and 75% ow rate conditions. In addition, the air ow rate was specied at 2000 m3 h1.
The thermal load of each zone was strongly dependent on the
OA temperature, occupancy density, heating load from the lighting
and equipment, set point SA-mass ow rate (msp,SA) in the DOAS,
and set point air temperature (Tsp) in the VRF system. The operation
schedules of the occupancy, lighting, and equipment were determined by an on-the-spot survey by the previous study [18] in the
same building. In addition, the air change rate for building inltration was assumed to be 0.732 h1, which was estimated based on
the volume of the building space and the capacity of the exhaust
fans in toilets [19]. Fig. 3 shows the temperature and humidity ratio
of the OA in the design day, and Fig. 4 shows the variation of occupancy density in each zone. msp,SA was estimated to satisfy IAQ,
which was represented by CO2 concentration. The CO2 concentration was dependent on the occupancy density and msp,SA. The CO2
concentration of the OA was assumed to be constant with 400 ppm.
3. Development of multi-objective optimal control strategy
3.1. Computational model for multi-objective optimal control
The energy performance of the VRF combined system installed
in a research ofce building was simulated using the EnergyPlus
199
Table 2
Specications of the VRF system and DOAS.
System
Details
0.0028
0.0024
-2
0.0020
-4
0.0016
-6
0.0012
0.0008
-8
0
10
12
14
16
18
20
22
model [20]. The VRF system was simulated using Raustad's model
[21] with the manufacturer's data. The heating performance curves
for the VRF system are given in Table 3. The capacity and power
consumption of the VRF system are expressed in terms of the
24
Time (h)
Fig. 3. Variation of outdoor air temperature and humidity ratio on the test day.
200
Table 3
Heating performance curves for the VRF system.
ODU
Curve
input
Piping
description
correction
factor
X
Y
Z
a
b
c
Lp
e
Normalized
factor
1.006
1.304E-3
3.699E-6
7.000E-9
e
f
g
X min
X max
Y min
Y max
e
e
e
7.5
225
e
e
IDU
Capacity Capacity ratio
modier
ratio
boundary function for low
T
EIR
EIR modier EIR modier EIR modier EIR modier
Capacity ratio
boundary function for function for function for function for
modier
high PLR
low PLR
high T
low T
function for high
T
Combination
ratio EIR
correction
factor
Capacity ratio
modier
function for T
TIDB
e
TOWB
TIDB
TOWB
Normalized
factor
1.170
4.903E-3
2.815E-4
TIDB
TOWB
Normalized
factor
2.044
6.447E-2
7.594E-4
TIDB
TOWB
Normalized
factor
29.73
2.074
1.817
0.1105
3.280E- 1.997E-3
2
1.713E-3 e
2.504E-2
3.888E-7
2.834E-4
16
24
20
2.2
5.335E-5
9.665E-7
2.138E-6
16
24
10
13.7
e
e
e
16
24
e
e
259.5
35.04
1.573
2.428E2
e
e
e
16
24
e
e
TIDB
e
TOWB
1.828E-2
8.816E-4
2.863E-3
16
24
20
0.7
TIDB
TOWB
Normalized
factor
0.8193
1.623E-3
6.109E-4
PLR
e
Normalized
factor
0.5309
1.144
2.532
PLR
e
Normalized
factor
1.221E3
3.304E3
2.969E3
CR
e
Normalized
factor
9.115
23.16
17.46
TIDB
TOWB
Normalized
factor
0.9646
1.029E-2
6.118E-4
0.9199
8.856E2
4.417
3.616E-2
5.779E-4
2.152E-3
16
24
7.6
13.7
e
e
e
0.5565
1
e
e
e
e
e
1
1.201
e
e
e
e
e
1
1.4
e
e
5.335E-5
9.665E-7
2.138E-6
16
24
20
13.7
Note: Z a bX cX2 dX3(for cubic curves); Z a bX cX2 eY fY2 gXY (for bi-quadratic curves).
increasing msp,SA, RHsp,SA, and Tsp, but the total energy consumption
is also substantially increased. Therefore, it is necessary to control
THC and IAQ simultaneously using the multi-objective optimal
control strategy, while minimizing the total energy consumption.
THC can be expressed by the PMV (predicted mean vote) [27,28]
and RH. The common ranges of winter comfort zone
are 0.5 PMV 0.5 and 30% RH 60% [27,29,30]. In this study,
the lower limit of PMV was extended to 1.0 in order to reduce the
total energy consumption according to Korean design standards
[31], which yielded the extended winter comfort zone. In addition,
an 8-hour average CO2 concentration must be lower than 1000 ppm
[32,33] to satisfy IAQ. Finally, the objective functions of the THC,
IAQ, and total energy consumption are expressed by Eqs. (1)e(3),
respectively.
ObjTHC
300
i1
Measured
Predicted
Energy consumption (GJ)
7
1X
Max0; PMVi 0:5 Max0; 1:0 PMVi
7
0:5 1:0
Cv(RMSE) = 7.6%
MBE = -4.1%
7
1X
Max0; RHi 60 Max0; 30 RHi
7
60 30
i1
(1)
200
ObjIAQ
(2)
i1
100
ObjE
0
7
1X
Max0; Ci 1000 Max0; 400 Ci
7
1000 400
.
l.
Ju Aug
Fig. 5. Comparison between the measured and the predicted monthly energy
consumption.
X
(3)
ObjTotal wfTHC $
ObjTHC ObjTHC;min
ObjTHC;max ObjTHC;min
ObjIAQ ObjIAQ ;min
wfIAQ $
ObjIAQ ;max ObjIAQ ;min
ObjE ObjE;min
wfE $
ObjE;max ObjE;min
MinObjtotal
0 kg s1 m_ SA 1:77 kg s1
30% RHSA 80%
20 C Tsp 24 Conly for control strategy B
(4)
(5)
In Eq. (5), the upper limit of msp,SA was determined by the peak
ventilation load, and the maximum RHsp,SA was determined to avoid
water condensation and mold growth [37].
3.3. Optimal weighting factors with response surface methodology
Since the objective functions were signicantly inuenced by
the weighting factors, the optimal weighting factors for the
control strategies A and B were determined by the RSM. The
relationships between the weighting factors and objective
functions were analyzed using the Box-Behnken design method.
As shown in Table 4, a total 13 runs were performed. As a result,
the second-order response surface model for each objective
function is expressed by Eq. (6). The coefcients in Eq. (6) are
given in Table 5. The coefcients of determination (R2) for
ObjTHC, ObjIAQ, ObjE, and ObjTotal were found to be 0.96, 0.92,
0.97, and 0.86 for the control strategy A, and 0.94, 0.89, 0.95,
and 0.99 for the control strategy B, respectively. Finally, the
optimal weighting factors for the THC, IAQ, and total energy
consumption were determined as 0.71, 0.64, and 1.00 for the
control strategy A, and 0.57, 0.69, and 0.72 for the control
strategy B, respectively.
2
2
Y b0 b1 wfTHC b2 wfIAQ b3 wfE b11 wfTHC
b22 wfIAQ
b33 wfE2 b12 wfTHC wfIAQ b13 wfTHC wfE b23 wfIAQ wfE
(6)
Table 4
Weighting factors assignment using the BoxeBehnken design method.
Case wfTHC, weighting
assigned to THC
wfIAQ, weighting
assigned to IAQ
1
2
3
4
5
6
7
8
9
10
11
12
13
0.0
0.5
0.0
0.0
1.0
1.0
0.5
0.5
0.5
1.0
0.0
1.0
0.5
0.0
0.0
0.5
0.5
0.5
0.5
0.0
1.0
1.0
0.0
1.0
1.0
0.5
0.5
0.0
0.0
1.0
0.0
1.0
1.0
0.0
1.0
0.5
0.5
0.5
0.5
201
202
Table 5
Coefcients of the second-order response surface model.
Coefcient
b0
b1
b2
b3
b11
b22
b33
b12
b13
b23
Strategy A
Strategy B
ObjTHC
ObjIAQ
ObjE
ObjTotal
ObjTHC
ObjIAQ
ObjE
ObjTotal
1.16
0.28
0.09
0.22
0.10
0.03
0.05
0.13
0.05
0.09
0.07
0.09
0.21
0.14
0.04
0.15
0.02
0.09
0.04
0.13
1,472,926
264,477
6846
389,010
32,450
21,162
112,581
123,663
6789
175,731
0.05
0.31
0.26
0.99
0.25
0.25
0.74
0.74
0.44
0.08
0.94
0.36
0.29
0.27
0.13
0.24
0.01
0.11
0.06
0.08
0.08
0.02
0.25
0.06
0.04
0.18
0.06
0.08
0.01
0.14
1,604,301
311,675
289,049
428,187
69,211
248,840
113,706
82,715
9088
134,198
0.17
0.46
0.43
0.55
0.34
0.24
0.30
0.16
0.60
0.12
Fig. 6. Variations of control variables in the TS strategy, DCVH strategy, optimal control
strategy A, and optimal control strategy B: (a) set point SA-mass ow rate and (b) set
point RH.
respectively. ObjTHC and ObjIAQ were relatively high at 1.72 and 0.27,
respectively, which indicates large deviation from the extended
winter comfort zone with poor IAQ. In the DCVH strategy, the daily
mean PMV and RH were improved from the TS strategy, which led
to better satisfaction for the THC and IAQ with ObjTHC, and ObjIAQ of
1.26 and 0, respectively. However, the DCVH strategy showed the
largest standard deviation of RH, which indicates severe humidity
imbalance between zones. Optimal control strategies A and B
showed better daily mean PMVs of 1.06 and 0.83, respectively,
compared with the TS and DCVH strategies. Optimal control
strategy B showed the lowest ObjTHC of 0.9 with a very low ObjIAQ of
0.004. However, optimal control strategy B showed higher total
energy consumption than the TS strategy and optimal control
strategy A due to the elevated set point air temperature. The DCVH
strategy showed the highest total energy consumption due to the
over-humidication and strict regulation for IAQ. The energy savings of optimal control strategies A and B against the DCVH strategy
were 29.0% and 20.4%, respectively. Overall, optimal control strategy B can be recommended as the best control strategy in the VRF
combined system because it yielded satisfactory THC and IAQ with
reasonable energy savings compared to the DCVH strategy.
Fig. 8 shows the variation of the CO2 concentration in Zone 3 for
the four strategies. Zone 3 was selected as an extreme IAQ condition
because it had the lowest IAQ with the highest occupancy density.
The variation of the CO2 concentration between zones was due to
the use of a central DOAS for ventilation in a multi-zone having
different occupancy densities. The 8-hour average CO2 concentrations for Zone 3 with the TS strategy, DCVH strategy, optimal control strategy A, and optimal control strategy B were 1704, 964, 1032,
and 998 ppm, respectively. Therefore, the TS strategy and optimal
control strategy A did not satisfy the IAQ guideline [32,33] in Zone 3
in terms of the 8-hour average CO2 concentration. The TS strategy
showed very high CO2 concentration, but it represented lower CO2
concentration below 1000 ppm during DOAS operations (9:00 AM,
13:00, 17:00, and 21:00 PM). The CO2 concentration for the DCVH
strategy was always controlled below 1000 ppm because msp,SA was
adjusted only for IAQ. Optimal control strategy B showed a slightly
higher CO2 concentration than the DCVH strategy, because msp,SA
was controlled for the multi-objective functions. In optimal control
strategy B, as the occupancy density increased rapidly (9:00 AM
and 19:00 PM) and the OA humidity decreased sharply (15:30 PM),
Table 6
Energy components for the four control strategies.
Energy components
TS strategy
DCVH strategy
Optimal strategy A
Optimal strategy B
EDOAS (MJ)
EHumidier (MJ)
EVRF (MJ)
28
e
1039
101
513
973
52
298
637
53
323
770
203
3000
TS strategy
DCVH strategy
Optimal strategy A
Optimal strategy B
2500
2000
1500
1000
500
0
10
12
14
16
18
20
22
24
Time (h)
Fig. 8. Variation of CO2 concentration in Zone 3 for the four control strategies.
5. Conclusions
Fig. 7. Indoor thermal conditions for (a) the TS strategy and DCVH strategy, and (b)
optimal control strategies A and B.
Table 7
Environmental performance and objective functions for the four control strategies.
Environmental performance and objective function
Environmental performance
Daily mean of PMV
Daily mean of RH (%)
Standard deviation of RH (%)
Ratio of the winter comfort zone over the total occupied hour (%)
Objective function
THC
IAQ
Energy consumption (MJ)
TS strategy
DCVH strategy
Optimal strategy A
1.38
20.7
4.9
5.6
1.26
43.1
9.5
35.9
1.06
33.0
8.5
47.9
1.72
0.27
1371
1.26
0
1801
1.13
0.01
1279
Optimal strategy B
0.83
33.0
7.4
55.0
0.90
0.004
1433
204
[9] Aynur TN, Hwang Y, Radermacher R. Simulation comparison of VAV and VRF
air conditioning systems in an existing building for the cooling season. Energy
Build 2009;41:1143e50.
[10] Aynur TN, Hwang Y, Radermacher R. Integration of variable refrigerant ow
and heat pump desiccant systems for the cooling season. Appl Therm Eng
2010;30:917e27.
[11] Goetzler W. Variable refrigerant ow systems. ASHRAE J 2007;49:24e31.
[12] Li YM, Wu JY. Energy simulation and analysis of the heat recovery variable
refrigerant ow system in winter. Energy Build 2010;42:1093e9.
[13] Aynur TN, Hwang Y, Radermacher R. Experimental evaluation of the ventilation effect on the performance of a VRV system in cooling modedpart I:
experimental evaluation. HVAC&R Res 2008;14:615e30.
[14] Zhu Y, Jin X, Fang X, Du Z. Optimal control of combined air conditioning
system with variable refrigerant ow and variable air volume for energy
saving. Int J Refrig 2014;42:14e25.
[15] Aynur TN, Hwang Y, Radermacher R. Simulation evaluation of the ventilation
effect on the performance of a VRV System in cooling modedpart II, simulation evaluation. HVAC&R Res 2008;14:783e95.
[16] Zhu Y, Jin X, Du Z, Fan B, Fang X. Simulation of variable refrigerant ow air
conditioning system in heating mode combined with outdoor air processing
unit. Energy Build 2014;68:571e9.
[17] Joo Y, Kang H, Ahn JH, Lee M, Kim Y. Performance characteristics of a simultaneous cooling and heating multi-heat pump at partial load conditions. Int J
Refrig 2011;34:893e901.
[18] Park DY, Kim KS. Comparison of actual building energy consumption with
simulated building energy consumption and economic analysis between VRF
ASHP and GSHP. J Archit Inst KOREA Plan Des 2014;30:241e8.
[19] Jeon J, Lee S, Hong D, Kim Y. Performance evaluation and modeling of a hybrid
cooling system combining a screw water chiller with a ground source heat
pump in a building. Energy 2010;35:2006e12.
[20] ENERGYPLUS. EnergyPlus engineering reference. Lawrence Berkeley National
Laboratory; 2013.
[21] Raustad R. A variable refrigerant ow heat pump computer model in EnergyPlus. ASHRAE Trans 2013.
[22] Liu X, Hong T. Comparison of energy efciency between variable refrigerant
ow systems and ground source heat pump systems. Energy Build 2010;42:
584e9.
[23] Wetter M. Simulation model air-to-air plate heat exchanger. Lawrence Berkeley Natl Lab; 1999.
[24] Wang S, Jin X. Model-based optimal control of VAV air-conditioning system
using genetic algorithm. Build Environ 2000;35:471e87.
[25] Kim DW, Park CS. Comparative control strategies of exterior and interior blind
systems. Light Res Technol 2012;44:291e308.
[26] M&V Guidelines: measurement and verication for federal energy projects.
version 3. US Department of Energy; 2008.
[27] ANSI/ASHRAE. ASHRAE standard 55e2004. Thermal Environmental Conditions for Human Occupancy. 2004.
[28] Olesen BW, Parsons KC. Introduction to thermal comfort standards and to the
proposed new version of EN ISO 7730. Energy Build 2002;34:537e48.
[29] ISO. 7730: ergonomics of the thermal environmenteAnalytical determination
and interpretation of thermal comfort using calculation of the PMV and PPD
indices and local thermal comfort criteria. 2005.
[30] ANSI/ASHRAE. ASHRAE standard 62e2001. Ventilation for Acceptable Indoor
Air Quality. 2001.
[31] SAREK. SAREK handbook. 3rd ed. Seoul: The Society of Air-Conditioning and
Refrigerating Engineers of Korea; 2011.
[32] Jeong JW, Choi A, No ST. Improvement in demand-controlled ventilation
simulation on multi-purposed facilities under an occupant based ventilation
standard. Simul Model Pract Theory 2010;18:51e62.
[33] Wong LT, Mui KW, Hui PS. A statistical model for characterizing common air
pollutants in air-conditioned ofces. Atmos Environ 2006;40:4246e57.
[34] Zeleny M, Cochrane JL. Multiple criteria decision making. McGraw-Hill New
York; 1982.
[35] Steuer RE. Multiple criteria optimization: theory, computation, and applications. Wiley; 1986.
[36] Miettinen K. Some methods for nonlinear multi-objective optimization. In:
Zitzler E, Thiele L, Deb K, Coello Coello C, Corne D, editors. Evol. multi-criterion
optim. SE - 1, vol. 1993. Springer Berlin Heidelberg; 2001. p. 1e20.
[37] Johansson P, Ekstrand-Tobin A, Svensson T, Bok G. Laboratory study to
determine the critical moisture level for mould growth on building materials.
Int Biodeterior Biodegrad 2012;73:23e32.
[38] ENERGYPLUS. EnergyPlus input output reference. Lawrence Berkeley National
Laboratory; 2013.