2023 Quantitative Framework Score IAQ

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Building and Environment 243 (2023) 110669

Contents lists available at ScienceDirect

Building and Environment


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

A novel quantitative assessment framework of the IAQ resilience


performance of buildings: The resilience score metric
Douaa Al Assaad *, Abantika Sengupta , Hilde Breesch
KU Leuven, Department of Civil Engineering, Building Physics and Sustainable Design, Ghent and Aalst Technology Campuses, Gebroeders De Smetstraat 1, 9000, Gent,
Belgium

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

Keywords: Resilience performance in building design has emerged as a critical consideration in the face of increasing un­
IAQ resilience certainties or ‘shocks’ posed by natural disasters, climate change & excessive pollution. To evaluate and optimize
Quantitative assessment framework building design decisions, holistic performance metrics are needed. This work defines a novel quantitative
Resilience score
assessment framework of indoor air quality (IAQ) resilience which output is the resilience score (RS) metric that
Degree of shock
Indoor air quality
integrates all resilience aspects and building-relevant pollutants. The framework was demonstrated via simula­
tions on a validated model of a case study educational building in Belgium for 3 ventilations systems: constant air
volume (CAV), demand-controlled ventilation (DCV), DCV without filters, and 3 shock types (mechanical MS,
internal IS, outdoor shocks OS). Results showed that OS was the least critical shock type, followed by IS and MS.
With increasing degree of shock, RSMS decreased for all 3 systems linearly by 64% until 2.8 h of power outage,
beyond which the rate of decrease slowed down considerably. CAV & DCV had the same RSMS and that of DCV
w/o filters was 13% lower. The RSIS for CAV & DCV deteriorated by 38.8% & 46% before plateauing at values of
0.45 & 0.36 respectively. CAV had better resilience against IS. The effect of filters did not reflect in the RSIS. The
RSOS for CAV & DCV deteriorated by 51% & 26% before plateauing at values of 0.49 & 0.73 respectively. DCV
had the best resilience against OS, followed by DCV w/o filters & CAV.

withstand its effects within an acceptable degradation range, and


1. Introduction recover from it back to its designed performance within a suitable time”
[9]. There has been much discussion and research regarding resilience
Over the past decades, the concept of “resilience” has gained assessment frameworks. These frameworks were either qualitative or
attraction among researchers and practitioners in organizational [1], quantitative and were motivated by the need to support design, decision
economic [2], social [3], and engineering domains (e.g., technical sys­ making, guide planning and eventually inform policy makers, which
tems, built environment, structures) [4]. While classic risk management enables resilience to be considered as an essential performance char­
tries to reduce the likelihood of disruptive scenarios and their conse­ acteristic in a system [10]. Qualitative frameworks offer a conceptual
quences on wellbeing of societies, resilience assessment aims to evaluate insight into resilience. They break it down into a set of characteristics or
the risks associated with undesirable events that cannot be predicted nor aspects, specific to the studied domain, identifying barriers, dependent
prevented; thus, denoted as “shocks”. Examples include power outages factors, and outputs allowing for risk-informed planning [11]. Recently,
and blackouts [5], natural disasters (e.g., earthquakes, floods) [6] or efforts have been made to conceptualize the resilience of the built
man-made catastrophes (e.g., bomb attacks, fires) [7], climate change environment against shocks that have detrimental consequences on the
(e.g., heatwaves, cold snaps) and excessive pollution (e.g., sudden sys­ wellbeing of occupants. This includes the framework of: (i) thermal
tem failures, smog, wildfires) [8]. Such events can disrupt the welfare of resilience against extreme overheating (due to heatwaves, power cuts)
populations for prolonged periods of time. This makes resilience an that threaten the lives of occupants by causing heat stress and impacting
important aspect to consider when designing and conceptualizing sys­ basic bodily functions [12] and (ii) IAQ resilience against excessive in­
tems and processes. door pollution risks causing acute exposure to multiple hazardous air
Across disciplines, resilience was defined as “a performance char­ pollutants, increasing the burden of disease [13,14]. Zhang et al. [15]
acteristic that describes the extent to which a system can absorb a shock, identified the four thermal resilience characteristics of cooling systems

* Corresponding author.
E-mail address: douaa.al-assaad@kuleuven.be (D. Al Assaad).

https://doi.org/10.1016/j.buildenv.2023.110669
Received 23 June 2023; Received in revised form 24 July 2023; Accepted 26 July 2023
Available online 27 July 2023
0360-1323/© 2023 Elsevier Ltd. All rights reserved.
D. Al Assaad et al. Building and Environment 243 (2023) 110669

Nomenclature kdecay Particle decay rate (h− 1)


MS Mechanical shock
AHU Air handling unit OS Outdoor shock
CAV Constant air volume ventilation system PM Particulate matter
Co Initial PM concentration (ppm) Q̇a Total airflow rate supplied by the AHU (m3/s)
CHAP Arithmetic mean of median concentrations of HAPs in each Q̇a,req Required air flow rate supplied by the AHU (m3/s)
zone (ppm)
Q̇a,z Airflow rate supplied to a zone z (m3s)
Cp,oa Ambient outdoor air concentration of species p (ppm)
RS Resilience score (− )
Cp,oa,expected Expected ambient outdoor air concentration of species p
RSCO2 CO2 resilience score (− )
(ppm)
RSHAPs HAPs resilience score (− )
Cp,sup Concentration of species p in supply air (ppm)
t time (h)
Cp,sup,expected Expected concentration of species p in supply air (ppm)
tocc Total occupied hours in one day (h)
Cp,z Concentration of species p in a zone z (ppm) TRV Threshold reference value (ppm)
CPM Concentration of particulate matter (ppm) tshock Shock duration (h)
DCV Demand controlled ventilation system VOCs Volatile organic compounds
doS Degree of shock (− )
doMS Degree of mechanical shock (− ) Greek symbols
doIS Degree of internal shock (− ) φ IAQ influencing variable (− )
doOS Degree of outdoor shock (− ) ε Effectiveness
Δtabs Absorptivity time (h) Δ Difference
Δtrec Recovery time (h) ω Weighting factor
ERp Actual emission rates of species p in each classroom (kg/h)
Subscripts
ERp,expected Expected emission rates of species p in each classroom
a air
(kg/h)
i HAP index
εabs Absorptivity effectiveness (− )
oa outdoor air
εrec Recovery effectiveness (− )
occ occupied
εres Resilience effectiveness (− )
p pollutants (CO2, VOCs, PM)
HAP Hazardous air pollutant
ref reference case
HQ Hazard quotient (− )
sup supplied
IAQ Indoor air quality
z zone
IDEAS Integrated district energy assessment by simulation
IS Internal shock

in buildings i.e., absorptive, adaptive, restorative capacity and recovery index was calculated as the difference in recovery capacities with and
speed. The resilience capacities can be evaluated into 3 categories: high without interventions. In the framework of thermal resilience, using a
(i.e., can increase or maintain its cooling capacity during shock), mod­ deterministic static approach, Ji et al. [20] proposed the thermal resil­
erate (i.e., can only maintain its cooling capacity during shock), and low ience index to quantify the resilience performance of buildings during
(i.e., decrease in cooling capacity). Recovery speed was classified as heatwave shocks. The unmet cooling ◦ C.hours index was extracted from
either high (i.e., system recovers within several hours), and low (i.e., the temporal temperature profiles and penalized with weighting factors
system recovery takes one or more days). Al Assaad et al. [16] identified based on degree of violations of standard effective temperature thresh­
three IAQ resilience aspects for indoor air quality (IAQ) management olds. The index was calculated for multiple cooling strategies and was
systems i.e., absorptive and recovery times and shock impact. used to classify and rate them into different classes. A similar approach
While qualitative frameworks are useful in gaining a deeper under­ was adopted by Homaei et al. [21] for cold snap events. In the same
standing into resilience, they do not quantify resilience aspects. Thus, framework and using a similar type of performance-based approach,
they cannot be applied to all disciplines where quantification of conse­ Sengupta et al. [22] classified and quantified shocks causing overheating
quences is essential to gain understanding of the shock scale. Semi- using the normalized degree of shock (doS). This approach allows to
quantitative approaches improved conceptual frameworks by assessing compare shocks and rate cooling strategies based on their impact on the
resilience characteristics on individual scales (0–100, 0–10) based on thermal environment through determining the total ◦ C.hours and occu­
expert observations or surveys, and combining them all into one index. pied hours above acceptable thresholds. Other metrics based on
Pettit et al. [17] measured vulnerability and capabilities of industrial indoor-outdoor temperature gradients were also developed [23–26]. In
supply chains against shocks, by conducting surveys targeted at policy the framework of IAQ resilience, Al Assaad et al. [16], classified shocks
makers. A weighted-sum approach based on experts’ opinions was into mechanical, internal, an outdoor and quantified them using the
conducted to aggregate the two resilience factors. normalized doS combining severity and duration. The degree of impact
A further improvement on semi-quantitative frameworks are fully of the shock was quantified using the total ppm.hours above acceptable
quantitative approaches that measure the system performance during a IAQ thresholds and used to compare smart and conventional ventilation
shock and benchmark it with respect to the design system performance systems in educational buildings.
during normal operation. Quantitative frameworks can be deterministic While existing quantitative frameworks can measure and thus eval­
or probabilistic, neglecting or considering the probability of shock uate the resilience level of a system – especially when resulting metrics
occurrence. They can also be either dynamic or static, accounting for are bounded, they are often used to quantify only one aspect of resilience
time dependence in resilience aspects [18]. Using a deterministic static (recovery capacity [19], degree of shock impact [16,20,22]). However,
approach, Rose [19] calculated the resilience of economies through the as agreed upon by multiple disciplines [4,16,17,23], resilience is a
0 to 1 ratio of expected over worst-case degraded performance levels. A multi-faceted construct with interdependent aspects (e.g., absorptivity,
dynamic approach was also developed where the dynamic resilience recovery, and degree of shock impact), each having a different type of

2
D. Al Assaad et al. Building and Environment 243 (2023) 110669

influence on the system performance and in some cases, on human under both normal and extreme shock conditions.
wellbeing. Consequently, to get a full insight on the resilience perfor­ The manuscript will be structured as follows: Section 2 showcases
mance of systems, there is a need for a holistic resilience quantitative the full framework and details the different steps. Section 3 presents the
assessment framework that combines and visualizes simultaneously the case study building, corresponding validated model in Modelica, and the
different identified aspects of resilience. simulation scenarios conducted to demonstrate the framework. Section
The current presented work will build upon previous research con­ 4 discusses the results of applying the framework on the case study.
ducted by the authors [16], where the concept of “IAQ resilience” was Section 5 lists the scope of RS and applicability in other disciplines as
introduced for the first time. The aim of the present work is to develop a well as its limitations and directions for future research. Section 6
holistic quantitative assessment framework of IAQ resilience against concludes the work and its main contributions.
different shock types that cause excessive pollution in buildings equip­
ped with any ventilation system(s) or IAQ management strategy. The 2. Resilience assessment framework
output of this framework is the resilience score (RS) – a bounded metric
that varies between 0 (i.e., worst resilience) and 1 (i.e., best resilience) Fig. 1 illustrates the proposed framework and the different steps to
and accounts simultaneously for the different aspects of IAQ resilience follow to assess the IAQ resilience of any building typology. This
defined in Ref. [16]. The RS accounts for different contaminants (CO2 assessment framework can be conducted experimentally on a case study
and hazardous air pollutants (HAPs): volatile organic compounds building, or virtually via a simulation model. Step 1 consists of inflicting
(VOCs) and particulate matter (PM)) weighted objectively based on either virtually or experimentally, a “shock” that causes a rise in the
combined quantitative and qualitative evidence of disease. The frame­ concentrations of prominent pollutants in the built environment for a
work will be demonstrated on a virtual model of a real-use educational certain duration. The shocks were expressed in terms of the degree of
building. This case study is the same one used by the authors in shock doS introduced in Ref. [16] and explained below. In Step 2, the
Ref. [16], where the corresponding model was validated experimentally experimentally measured or numerically predicted temporal

Fig. 1. Quantitative assessment framework of IAQ resilience performance of any building typology.

3
D. Al Assaad et al. Building and Environment 243 (2023) 110669

concentration of pollutants are extracted from the building zones. In in Ref. [16] that pollutants’ concentrations built-up quite fast (i.e., in the
Step 3, the normalized indices of the three aspects of IAQ resilience range of minutes) causing acute exposure events, which have detri­
(absorptivity effectiveness εabs , recovery effectiveness εrec and degree of mental effects on human health that extend well after the shock is over.
shock impact εres ) defined in Ref. [16] and explained below are calcu­ To assess IAQ resilience, Al Assaad et al. [16] classified shocks into 3
lated from the concentration profiles. They are averaged across the types and quantified them by introducing the normalized degree of
building zones (to obtain the mean value for the building), for each shock (doS) (equation 1, Table 1) combining the shock severity (i.e., a
pollutant across the doS range. In Step 4, the calculated indices repre­ measure of deviation of an IAQ related variable φactual from expected
senting the three resilience aspects, are visualized on a spider chart for conditions φexpected leading to a significant decline in IAQ) and the shock
the different building-relevant contaminants across the doS range. The duration (tshock ) normalized by an observation period: the occupied
RS for each pollutant is then computed as the area of the formed triangle period (tocc ). The higher the doS, the worse the IAQ degradation.
in the spider chart. To calculate one single RS representative of the The 3 types of shock [16] were summarized in Table 1. Note that the
building and accounting simultaneously for all the pollutants, the doMS, doIS and doOS defined in Table 1 do not consider spatial effects
weighted sum of the individual RS for the pollutants is computed across due to non-homogeneity of the airflow field and associated pollutants’
the doS range (Step 7). The evolution of the resilience with increasing concentration fields. In this framework, it was assumed that the space is
shock degree can be analyzed. The weighting of the different RS is one control volume with lumped conditions, where doMS, doIS and doOS
conducted in Steps 5 and 6, objectively via a combined quantitative and were uniform in the space and were the same for all pollutants,
qualitative assessment based on exposure levels and evidence of harm regardless of their emission patterns.
and disease respectively, as will be explained below [27]. Section 2.1
explains the shocks and their quantification (Step 1). Section 2.2 defines
2.2. The three aspects of IAQ resilience
the aspects (or characteristics) of resilience and their indices (Steps 2, 3,
4). Section 2.3 presents the pollutant-based and aggregated
Fig. 2 illustrates the evolution of Cp (t) when a shock occurs at a time
building-level RS and the pollutants’ prioritization methodology to
to for multiple possible scenarios (1,2,3). Scenarios 1, 2 and 3 correspond
compute the weighting factors across the doS range (Steps 5, 6, 7).
to different systems undergoing the same shock having a degree doS or
to the same system undergoing different doS. These scenarios are pre­
sented to show how different systems can have different resilience re­
2.1. Shocks’ classification and quantification
sponses to the same shocks or how IAQ can deteriorate differently
depending on the doS. In the framework of IAQ resilience, three aspects
To test the resilience of a building, shocks should be inflicted (Step 1,
of resilience are defined: (1) absorptivity, (2) recovery and (3) impact.
Fig. 1). They are defined as extreme unexpected events that occur
These three aspects exist simultaneously, or in other words, one cannot
without prior knowledge of the building residents or designer and thus,
exist without the other.
no prior action (e.g., preventive maintenance) can be taken to prevent
them. Shocks should be realistic and thus, could have already occurred
• Resilience aspect 1: Absorptivity
or will occur in the future. In the framework of IAQ resilience, shocks
cause a significant degradation in performance of installed ventilation
The period between Cp (t) crossing the threshold reference value
systems or any other type of installation that manages IAQ. Shocks cause
a quick build-up of various pollutants generated outdoors and/or in­ (TRV) for pollutant p (t1 for scenario 1, t′1 for scenarios 2, t″1 for scenarios
doors from occupants, furniture, building envelope, to levels that can 3) until it reaches the maximum concentration Cmax at times t2 , t′2 , t″2 was
exceed acceptable toxicological reference values (TRV) set by standards. denoted as absorptivity time (Fig. 2). In the specific case where, Cp (t)
Fig. 2 illustrates the temporal evolution of the concentration of a was already above the TRV when the shock occurred (i.e., base case
pollutant p, Cp (t) during shock events for multiple scenarios. It was seen concentrations above the threshold), then absorptivity time is the

Fig. 2. Simplified illustration of the temporal evolution of the concentration of a pollutant p during shock events.

4
D. Al Assaad et al. Building and Environment 243 (2023) 110669

Table 1 performance is proportional to absorptivity times. The normalized


Illustration of the three types of shock (MS, IS and OS) and their quantification absorptivity effectiveness index was defined below as:
using the degree of shock (doS) based on the previous work [16].
Δtabs
Type of shock Description Quantification (degree of shock, doS) εabs = (6)
φexpected − φactual tshock
tocc
doS = × (1)
φexpected tocc
⏟̅̅̅̅̅̅̅̅̅̅̅̅̅⏞⏞̅̅̅̅̅̅̅̅̅̅̅̅̅⏟ ⏟̅⏞⏞̅⏟
Duration
Where Δtabs is the absorptivity time (h) benchmarked with respect to the
occupied period tocc (h). tocc was chosen since any absorptivity time
Severity

Mechanical They occur due to Q̇a,req,z − Q̇a,z tshock


doMS = (2) lasting beyond the occupied period is not relevant for the resilience
shock partial or complete Q̇a,req,z
×
tocc
(MS) failure of system ⏟̅̅̅̅̅̅̅̅̅̅⏞⏞̅̅̅̅̅̅̅̅̅̅⏟ ⏟̅⏞⏞̅⏟ performance since occupants were no longer residing in the space. εabs
Duration
component(s). If the MS
Severity
varies between 0 (i.e., worst case scenario, the system absorbs the shock
Cp,sup − Cp,sup expected tshock
affects ventilation doMSp =
Cp,sup expected
×
tocc
quickly) and 1 (i.e., best case scenario, the system takes considerable
system components, it ⏟̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅⏞⏞̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅⏟ ⏟̅⏞⏞̅⏟ amount of time to fully absorb the shock). Note that if the system did not
Duration
leads to a decrease or Severity
(3) react to the occurrence of a shock (i.e., no recorded absorptivity time)
interruption in the
supply of clean outdoor does not mean that Δtabs = 0, bur rather that equation (6) cannot be
air to a zone z applied.
(Q̇a,z < Q̇a,req,z ) (e.g.,
faulty fans, dampers • Resilience aspect 2: Recovery
stuck in closed position,
filter fowling [28],
leaking ducts, power The period between the shock end time t3 or when an intervention
outages Q̇a,z = 0 [29]). strategy became effective such that Cp (t) decreased significantly from its
If the MS affects the peak Cmax to reach the TRV or lower (t4 for scenarios, t′4 for scenarios 2,
components of other
types of systems that
t″4 for scenarios 3, Fig. 2), was denoted as recovery time. In the specific
regulate the case where, Cp (t) was already above the TRV in the base case of no
concentrations of shock, then the recovery time is the period between t3 and the moment
pollutants in the supply Cp (t) rejoined the base case scenario concentration profile.
air (e.g., air cleaners), it
leads to an increase in
Different systems could have different recovery times and thus
the pollutants’ different resilience performance. The quicker that the system and indoor
concentration in the space recovered from the shock, the faster the occupant exposure sub­
supply air sided. The slower the system recovery time, the longer occupants were
(Cp,sup > Cp,sup expected )
left to inhale concentrations of pollutants that were higher than the TRV
Indoor They occur inside the ERp − ERp,expected tshock
shock (IS) zones and are linked to
doISp =
ERp,expected
×
tocc
(4) (Fig. 2). Thus, the resilience performance was inversely proportional to
occupancy. They consist
⏟̅̅̅̅̅̅̅̅̅̅̅̅̅⏞⏞̅̅̅̅̅̅̅̅̅̅̅̅̅⏟ ⏟̅⏞⏞̅⏟
Severity Duration recovery times. The normalized recovery effectiveness index was
of an increased defined below as:
generation rate of
pollutants beyond the Δtrec
εrec = 1 − (7)
capacity of the systems tocc
(ERp > ERp,expected ) (e.g.,
sick occupants Where Δtrec is the recovery time (h) benchmarked with respect to the
generating pathogens,
occupied period tocc (h). tocc was chosen since any recovery time lasting
several additional
healthy occupants beyond the occupied period was not relevant for the resilience perfor­
entering the room [30]) mance since occupants were no longer residing in the space. εrec varies
Outdoor They occur due to
doOSp =
Cp,oa − Cp,oa expected
×
tshock
(5) between 0 (i.e., worst case scenario, the system recovers slowly
shock (OS) outdoor pollution Cp,oa expected tocc throughout the occupied period) and 1 (i.e., best case scenario, the
⏟̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅⏞⏞̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅⏟ ⏟̅⏞⏞̅⏟
events, where generated
pollutants can infiltrate
Severity Duration
system recovers almost instantaneously). Note that if the system did not
through the envelope or react to the occurrence of a shock (i.e., no recorded absorptivity and thus
the ventilation system no recovery) does not mean that Δtrec = 0, bur rather that equation (7)
(Cp,oa > Cp,oa expected ) (e. cannot be applied.
g., traffic jams, fire
hazards, smog [31])
• Resilience aspect 3: Shock impact

period between Cp (t) deviating from what would be the base case The impact of the shock is the unmet ventilation ppm.hours above the
concentrations with no shock until it reached the maximum concen­ TRV (Fig. 2). The ppm.hours is a measure of occupant exposure to con­
tration Cmax . Cmax is the saturation concentration (i.e., the highest con­ centrations exceeding the TRV during a certain period, which in this
centration that can be reached in that space for that doS. For t > t2 , t′2 , t″2 case is the occupied period tocc (h). Different systems having different
until t3 (end of shock event) is called the adaptation period, wherein ventilation effectiveness could yield different impacts (Fig. 2). The shock
the systems in place may take action to circumvent shock effects on IAQ. impact considers the adaptation time if it exists. The shock impact was
If the shock is short-term, and saturation concentration cannot be expressed via the normalized resilience effectiveness index below:
reached, the adaptation time can be zero (Fig. 2). ppm.hoursref − ppm.hourssystem under shock
Different systems could have different absorptivity times and thus εres = (8)
ppm.hoursref − ppm.hourssystem normal operation
different resilience performance. The quicker the absorptivity (i.e., low
absorptivity time), the quicker the build-up and IAQ deterioration and The ppm.hourssystem under shock of the evaluated system under a shockwas
the quicker Cp (t) reaches its maximal value Cmax . Thus, for low ab­ compared to two values: ppm.hourssystem normal operation in the case of no
sorptivity times, the acute exposure of occupants to pollutants is more shock and a reference ppm.hoursref . The ppm.hoursref is the unmet
severe since they inhale the same mass of pollutants but in a shorter time ventilation ppm.hours for a reference building scenario that yields the
and stay exposed to Cmax longer (Fig. 2). Thus, the resilience worst possible IAQ during normal operation and during a shock. This

5
D. Al Assaad et al. Building and Environment 243 (2023) 110669

reference case scenario is a building with no IAQ management strategy and implicitly considers absorptivity and recovery in its calculation, it is
(no ventilation system or any auxiliary). If the ppm.hourssystem under shock important to consider it independently to differentiate between systems:
approached the ppm.hourssystem normal operation , then the system and thus the For example, two systems can have the same absorptivity and recovery
indoor environment and occupants were barely affected by the shock, effectiveness (εabs , εrec ), but one system can have lower ppm.hours above
and kept an operation that is close to that of a base case scenario where the TRV and thus lower εres .
no shock occurred. In this case, εres approaches 1 (i.e., lowest impact and There are multiple ways this graph could be used. If the objective is
best resilience). However, if the ppm.hourssystem under shock approached the to compare multiple systems or test an intervention strategy, the spider
ppm.hoursref , then the system and thus the built environment approach chart could be plotted for the systems in questions for each doS in the
the case of a building with no ventilation or auxiliary systems with the range and for each pollutant p. If the objective is to gain understanding
worst IAQ (i.e., as if the system is not there and is incapable of any on how different pollutants react to shocks for one building scenario, the
remedy action). In this case, εres approaches 0 (i.e., biggest impact and chart can be plotted for multiple pollutants for each doS in the range.
worst resilience). The chart can be used to analyze and compare different shock types
Note that the three presented aspects of IAQ resilience (absorptiv­ having the same doS, for each pollutant p or even the same type of shock
ity, recovery, and impact) as seen on the concentration profile in Fig. 2 for multiple doS.
are counted as one resilience cycle [32]. If, during the shock, the built
environment recovered (i.e., due to occupancy fluctuations, interven­ 2.3. The resilience score (RS) metric
tion strategy or shock ends) and then re-absorbed the same or another
shock, another cycle starts. The absorptivity, recover times (Δtabs , Δtrec ) The total resilience score (RS) metric of a building is the normalized
and the degree of shock impact (ppm.hourssystem under shock ) are then sum of the RS of each pollutant p. The RS of each pollutant p is calcu­
re-calculated and added to the calculated values from earlier cycles. lated as the area of the triangle formed by εabs , εrec & εres in the spider-
The resilience cycle depends on factors other than the system type chart seen in Fig. 3. The total RS can be calculated in equation (9) as:
and its control, such as the type of shock (MS, IS, OS), the degree of ⎧ ⎫

⎪ ⎪

shock (doS), space characteristics (e.g., volume, envelope tightness, user ⎪ ⎪
5 ⎨ ∑
#HAPs ⎬
behavior), the type of pollutant in question and notably the occupancy RSCO2 + ωi RSi
13 ⎪ ⎪
schedule in the building, since occupants are important generators of RS = ⎪ ⎪
⎩ i=1
⏟̅̅̅̅̅̅̅⏞⏞̅̅̅̅̅̅̅⏟


⎭ (9)
indoor pollutants. Thus, when conducting the resilience assessment, it is RSHAPs

important to apply and report the results of the framework under the ∑
where ωi = 1
building’s most representative user profiles. In the case where the
building typology is characterized by more stochastic user profiles (e.g., The total RS splits the building pollutants between CO2 and haz­
residential buildings), multiple profiles should be tested, and a range of ardous air pollutants (HAPs) (e.g., particles, volatile organic compounds
values should be given for the different resilience aspects and thus the (VOCs), etc.) due to their different impacts on health and building de­
eventual RS. Moreover, in a building with different zones having signers’ understanding of them. CO2 if inhaled, does not cause disease to
different dimensions, indoor/outdoor pollution sources, emission humans like HAPs. However, it is an asphyxiant: Exposure to high levels
strengths, as well different user profiles, the framework (Fig. 1) should of CO2 compromises the immunity system [13], causes sick building
be applied to each zone individually. The range of RS as function of syndrome (i.e., nausea, fatigue, headaches, dizziness, drowsiness) [33,
increasing doS can be reported for the building. The arithmetic mean 34] and decreased cognitive performances [35] and at extremely high
values could be considered as a representative of the building. The levels (>10–20%), loss of consciousness and even death [36]. RSCO2
framework can be applied similarly to different building typologies (i.e., (equation (9)) is the score calculated for CO2, the pollutant considered to
residential, non-residential, or commercial). be an occupancy indicator [37] and relied upon by partitioners to design
After calculating the building zones’ average three metrics (εabs , εrec , and size the ventilation system [38].
εres ) (Steps 2 & 3) for each pollutant p, a spider chart could be plotted The weights associated with each HAP are calculated based on the
with these three metrics to visualize and gain understanding on the full recently developed method by Sérafin et al. [27] for prioritization of
picture of the IAQ resilience performance of the building and systems inhaled indoor air pollutants. This method is based on combined
under evaluation. This can be seen in Fig. 3 for the scenarios 1, 2, 3. The quantitative approach based on an exposure risk assessment and a
idea behind the spider charts is to juxtapose the aspects of resilience that qualitative approach based on hazard classification. The prioritization of
are approximately independent in their definition. The absorptivity and HAPs improves the identification of pollutants of concern for human
recovery times are independent one from the other as their associated health, which is important during a IAQ resilience assessment.
phenomena occur due to different events or system stimuli. Moreover,
they cannot occur simultaneously. The absorptivity is a response to the 2.3.1. HAPs’ prioritization: quantitative and qualitative approach
occurrence of shock and recovery is a response to the end of a shock. The Existing methods in the literature for HAPs’ prioritization were
degree of shock impact expressed with the ppm.hours spent above the based on toxicological data inputs (i.e., quantitative approaches). They
TRV reflects the cumulative health impact of the shock. While it partly ranked substances by scoring each based on the weighted sum of the sub-
scores of the risks related to different exposure routes, duration, and
different health effects [39–41]. While providing a clear ranking from
highest to lowest priority pollutant, the sub-scores and their weights
were calculated based on random decisions; making it difficult to
compare health effects. Moreover, they allowed for trade-offs between
health effects leading to ignore certain pollutants. Sérafin et al. [27]
proposed an improved new method based on combining risk assessment
from toxicological concentration data and qualitative approach based on
evidence of harm. This method was applied in this framework to
calculate the weights ωi (equation (9)) and was explained below:

• Quantitative risk assessment:


Fig. 3. Spider chart visualizing simultaneously the three aspects of IAQ resil­
ience (εabs , εrec , εres ). Scenarios 1, 2, 3 correspond to the ones found in Fig. 2.

6
D. Al Assaad et al. Building and Environment 243 (2023) 110669

The risk associated with inhalation exposure to HAPs was expressed HAPs from different international organizations (US EPA, the Interna­
by the calculation of the hazard quotient (HQ), which is the ratio of the tional Agency for research on Cancer (IARC), European Commission)
exposure concentration of the HAP to the TRV concentration at which no that were based on animal or human evidence of disease. The result of
adverse health effects occur: this combination was a 5-class hazard classification, the “CMRE” which
considers Carcinogenic risks, potential for Mutagenicity and Repro­
CHAP
HQ = (10) ductive toxicity, or Endocrine disruptions simultaneously. A substance
TRV
was a priority if it belongs to CMRE category 1 or 2 corresponding to
CHAP is the arithmetic mean of the median concentrations measured or substances that are known to have (category 1) or most likely to have
predicted in each building zone i during the observation period (Fig. 2). (category 2) at least one of the aforementioned health effects in humans.
The TRVs were extracted from recognized national or international or­ The rest of the categories were either unclassified substances or sub­
ganizations (e.g., US environmental protection agency (EPA) [42], stances with no proof of harm or no sufficient evidence. Their qualitative
world health organization (WHO) [43], Flemish guidelines [44], Euro­ assessment results showed 71 substances of priority including PM2.5,
pean Commission [45] and others [46,47]). Different TRVs may exist for PM10, VOCs (specifically formaldehyde, acetaldehyde, benzene,
the same substances, however, this is beyond the scope of this study. A dichloropropane) which fell into categories CMRE 1 or 2.
substance was a priority if its HQmean at building level or the 95th To use this method in the resilience assessment framework, a
percentile HQP95 was higher than 1. Note that in Ref. [27], HQ was not quantitative risk assessment was conducted for the substances that were
weighted with the occupied time in the zones. identified as priority pollutants by the CMRE classification [27]. The HQ
distribution of each HAP was calculated. The first occurrences of HQ > 1
• Qualitative risk assessment: of each HAP were used to weight the HAPs against each other and

calculate the weighting factors such that ωi = 1 (equation (9)) and to
Sérafin et al. [27] combined the different qualitative classification of calculate the total RS of the building. This was done so as not to neglect

Fig. 4. Illustration of the a) test lecture room building consisting of classrooms E120, E220 and the AHU, b) corresponding validated model in Modelica.

7
D. Al Assaad et al. Building and Environment 243 (2023) 110669

the weight of priority substances, which HQmedian does not always exceed The average outdoor concentration of PM2.5 and PM10 were obtained
1 but rather its 95th percentile. from an air quality monitoring station in Gent located in the vicinity of
the case study building [62]. The recorded outdoor concentrations of
3. Application of the framework PM2.5 and PM10 have a monthly average of 10 μ g/m3 and 18 μ g/m3
respectively. The validated filter model of Faulkner et al. [63] was used
3.1. Case study building and corresponding model to model the AHU G4 and E10 filters. Additionally, the PM decay model
due to gravitational settling expressed in equation (11) and validated in
The resilience assessment framework will be applied and demon­ Ref. [63] was used:
strated on a case study educational building located in the city of Gent, ( )
CPM (t) = Co exp − kdecay t (11)
Belgium at the Campus Gent of KU Leuven [16] consisting of 2 class­
rooms (E120 and E220) [48]. Both classrooms (13.28 m length × 10.69
Where CPM (t) is the transient PM concentration in the zones, Co is the
m width × 2.7 m height) can seat a maximum of 80 students each. The
initial PM concentration and kdecay (h− 1) is a defined constant rate of PM
test lecture room building is a real-use case study, where regular classes
decay that depends on the particle diameter size. It was equal to 0.39 h− 1
and exams are held throughout the academic year spanning from
for PM2.5 and 1.01 h− 1 for PM10 [64]. The TRV for PM2.5 and PM10 were
September until June. These schedules have been consistent throughout
equal to 15 μ g/m3 and 45 μ g/m3 according to WHO’s air quality
consecutive academic years and can thus be considered as representa­
guidelines also adopted in Belgium [65].
tive. Fig. 4a illustrates a schematic of the building and its AHU. Detailed
Note that other contaminants include outdoor generated pollutants
description of the building envelope can be found in Refs. [16,22].
such as NOx. However, they won’t be considered in this work as it was
The AHU supplied 100% conditioned clean filtered outdoor air with
seen in Ref. [16], that concentrations did not exceed recommended
a maximum airflow of 4400 m3/h and a minimum of 400 m3/h. The air
Flemish guideline values even under extreme OS events. According to
was first filtered by a G4 prefilter for coarse particles to protect the AHU
the CMRE classification [27], PM2.5, PM10 and the VOCs found in the test
and its components. It has a filtration efficiency of 15% for particles
lecture building, classify as CMRE ≤ 2 and will be ranked according to
having an optical diameter of 0.3–2.5 μm (PM2.5) and an efficiency of
their exposure to calculate the total RS.
70% for particles between 2.5 and 10 μm (PM2.5-10). This filter is fol­
The numerical predictions of the Modelica model were validated
lowed by an E10 filter having an efficiency of 80% across all particle
through experiments. The zone-average temporal variations of temper­
sizes. These filters were tested over a period of 14 months and exhibited
ature and species’ concentrations were validated for the case of normal
a stable behavior with minimal pressure drop [49].
operation and a case with a 30-min power outage during a full day.
The installed ventilation system is a balanced all-air mechanical
Detailed description of experiments can be found in Ref. [16].
ventilation applying the concept of demand-controlled ventilation
(DCV). Each classroom was a single zone, with supply and return vari­
able air volume boxes regulated by a request signal from a PI control 3.2. Simulation scenarios
block that regulated the airflow based on competing room temperature
and CO2 concentrations (Fig. 4a). The system can be switched to con­ IAQ resilience was assessed during an observation period of one day
stant air volume (CAV). The AHU was designed to maintain yearlong on Monday June the 6th. Fig. 5 illustrates the measured occupancy
constant temperatures in the classrooms of 22 ± 0.5 ◦ C and CO2 con­ schedules on that day. The generation of the occupancy schedules were
centrations of 1000 ppm. The AHU operated during weekdays from 7:30 explained in Ref. [16]. The average occupied period between both
a.m. to 6:00 p.m. and was shut off during weekends. The operation and classrooms was equal to 587 min.
control of both heating and cooling systems can be found in detail in MS, IS and OS were inflicted in the model (Step 1, Fig. 1). A power
Ref. [16]. outage was considered to represent a MS (complete interruption in
In [16], a model of the building, and its AHU were developed in supply of air). The duration of the power outage was varied from 0 to
Modelica [50]. Fig. 4b illustrates the model of the test lecture room 587 min (complete occupied period) with a 20-min increment (i.e.,
building as seen in Dymola as well as an overview of the components. doMS ∈ [0, 1], Δ(doMS) = 0.05, equation 2, Table 1). IS was simulated
The “Integrated District Energy Assessment by Simulation” (IDEAS) library with an additional number of occupants in the E220 beyond the AHU
[51] that contains validated sub-models of the building envelope, sys­ capacity throughout the occupied period. The total number varied be­
tems, and their control was used. The considered contaminants in the tween 58 (max capacity) to 116 occupants with an increment of 5 oc­
model (and relevant to the current building as informed by its building cupants (i.e., doISp ∈ [0, 1], Δ(doIS) = 0.05, equation 4, Table 1). OS was
monitoring system [52–54]) are: simulated by an outdoor pollution event (i.e., traffic jam in the early
1CO2 due to breathing activities with a generation rate of 0.0037 L/s. morning 6:00 a.m.–10:00 a.m.) that increased outdoor levels of pollu­
person [55]. The average outdoor concentrations were CCO2 ,oa = 450 tion from 0 to 2.5 × (i.e., doOSp ∈ [0, 1], Δ(doOS) = 0.05, equation (5),
ppm [56]. The TRV for CO2 was equal to CCO2 ,TRV = 1000 ppm according Table 1). To test the framework, the same simulations were also applied
to Flemish guidelines in Belgium [57]. to the case of CAV system operating at a constant air rate of 4400 m3/h
2VOCs consisting mainly of aldehydes as was experimentally and the same DCV system without the presence of filters in the AHU.
measured in the test lecture rooms [16]. For the occupants, a total VOCs’
generation rate of 1.73 μg/s.person was assumed [58]. As for the 4. Results and discussion
building envelope and furniture (tables), the generation rate was
determined experimentally in Ref. [16] and was equal to 7.8 μg/s. The The demonstration of the framework will be presented in the order of
average outdoor concentrations in Gent, were equal to CVOC,oa = 0.0068 the steps in Fig. 1. Section 4.1 presents the visualization of the resilience
ppm [59]. The TRV for VOCs was equal to CVOCs,TRV = 0.049 ppm [57]. aspects as function of the contaminants for the different ventilation
systems and shock types (Step 3, 4). Sections 4.2 and 4.3 present the
3. PM due to occupant activities and infiltration. Occupant activities ranking of the HAPs (Step 5, 6) and the RS (Step 7).
generating PM in classrooms involve students/teachers walking
around the room [60], sitting, and fidgeting. It was found that the 4.1. Resilience aspects for different shock types
source strengths for human activities ranged from 0.03 to 0.5
mg/min for PM2.5 and from 0.1 to 1.4 mg/min for PM10 [61]. In this Figs. 6 and 7 & 8 illustrate the spider charts grouping the different
study, the average of each range was considered. resilience aspects (εres , εabs , εrec ) segregated as function of the different
indoor contaminants (CO2, VOCs, PM2.5, PM10) for MS, IS & OS

8
D. Al Assaad et al. Building and Environment 243 (2023) 110669

Fig. 5. Measured occupancy schedules in a) E120 and b) E220.

respectively for a) CAV, b) DCV and c) DCV without filters in the AHU. TRV value as is the case with CAV.
The spider charts were presented for 3 ranges of doS (low, mid, and For CAV (Fig. 6a, Fig. S1), with increasing doMS from 0.288 to 1,
high). εres,CO2 decreased linearly from 0.94 to 0.28, εabs,CO2 increased linearly
Out of all shock types, MS was the only shock type that challenged from 0.15 to 0.53 plateauing at doMS = 0.721 until 1. εrec,CO2 was con­
the IAQ resilience of all systems with respect to all contaminants stant at 0.92 until doMS = 0.577 then decreased linearly to 0.62 at doMS
throughout the entire doMS range (Fig. 6). This was not the case for IS = 1. Note that there was a slight increase in εrec,CO2 at doMS = 0.577 as
and OS. was the case with VOCs due to the mid-day occupancy break, however,
IS caused violations of CO2 and VOCs TRVs for CAV throughout the not as pronounced as CO2 was a faster reacting specie. Similar trends and
entire doIS range (Fig. 7a), for DCV with/without filters (Fig. 7b and c) values were obtained for DCV as the system was driven by CO2 demand
for doIS > 0.22. The reason for the latter was that in DCV systems driven and thus behaved similarly to CAV (Fig. 6b).
solely by CO2 demand, VOCs’ concentrations tend to build-up during For both CAV & DCV (Fig. 6a, Fig. S1), with increasing doMS from
periods when no significant occupancy is present in the classrooms (i.e., 0.288 to 1, εres,PM2.5 decreased linearly from 0.97 to 0.3, εabs,PM2.5
early morning, evening, class breaks). Thus, it takes a more intense IS to increased linearly from 0.1 to 0.48 plateauing at doMS = 0.721 until
cause VOCs’ concentrations to deviate noticeably from the base case of doMS = 1. εrec,PM2.5 decreased a slow linear decrease from 0.97 to 0.91 at
No Shock. Note that for the DCV w/o filters (Fig. 7c), only PM2.5 TRVs doMS = 0.721 and then a faster linear decrease from 0.91 to 0.62 at
were violated for extreme IS (doIS = 1). Thus, IS do not challenge the doMS = 1. The same trends were obtained for CAV and DCV. CAV & DCV
resilience of mechanical ventilation systems when it comes to fine and had similar trends with PM10 as well with slightly higher εres,PM10 , lower
coarse PM except for when systems don’t have adequate filtration. OS εabs,PM10 and higher εrec,PM10 due to stronger gravitational settling effects.
caused violations of VOCs only in the case of a CAV system for doOS > Filters mainly reduced εres,PM2.5 , εres,PM10 and increasingly so with an
0.29 and not for DCV (Fig. 8a and b). It caused PM2.5 violations in the increasing doMS (e.g., εres,PM2.5 (doMS = 0. 432) = 0.83 for DCV & 0.8 for
case of a DCV system with no filters for doOS > 0.29 (Fig. 8c). DCV w/o filters, εres,PM2.5 (doMS = 1) = 0.3 for DCV & 0.15 for DCV w/o
Looking at Fig. 6, across the entire doMS range and for both CAV and filters). On the other hand, filters did not affect absorptivity and re­
DCV, VOCs had the lowest εres (highest shock impact), the highest εabs , covery responses. Despite the MS being a power outage (i.e., no venti­
(gradual absorption) and the lowest εrec , (slowest recovery). This was lation), the effect of filters was still there due to prior PM build-up at the
followed by CO2, PM2.5 and PM10. Recovery of PM was higher due to time of shock occurrence.
gravitational settling effects, which was more pronounced for PM10. Another observation with MS (Fig. S1), was the plateauing of the
For CAV (Fig. 6a, Fig. S1 in the supporting information file), with absorptivity effectiveness εabs for all contaminants and for all systems
increasing doMS from 0.144 to 1, εres,VOCs decreased linearly from 0.85 to starting doMS = 0.721, which occurred simultaneously with the fast
0.25. εabs,VOCs increased linearly from 0.13 to 0.65 plateauing at doMS = linear decrease in εrec . After 7 h of power outage, there was a lower
0.721. εrec,VOCs remained 0.84 until doMS = 0.432, then slightly occupancy for afternoon classes. Thus, the rate of generation of endog­
increased to 0.87 at doMS = 0.577. For doMS > 0.577 until 1, εrec,VOCs enous contaminants from respiratory & occupants’ activities decreased
decreased linearly to 0.54. This was due to the longer power outage. significantly over time.
Thus, the zone has more time to store VOCs and takes more time to Looking at Fig. 7, across the entire doIS range for CAV, VOCs and CO2
recover. The slight increase in εrec,VOCs at doMS = 0.577 is due to the had similar resilience aspects (εres , εabs , εrec ). For DCV, for doIS > 0.22,
reduction in occupancy at the time of shock events (i.e., mid-day break, VOCs and CO2 had equal resilience aspects with CO2 having slightly
Fig. 5) speeding up the recovery process. Similar trends were obtained lower εrec . In the case of DCV w/o filters, when PM2.5 challenged the
for DCV (Fig. 6b and c). The difference was that εres,VOCs was ∼ 6% lower resilience for doIS > 0.62, it had similar resilience aspects as CO2.
and εrec,VOCs was ∼ 7% higher. This shows that the DCV system has For CAV & DCV (Fig. 7, Fig. S2), with increasing doIS from 0.07 to 1,
slightly higher VOCs violations. However, recovery was slightly faster as εres,VOCs remained almost constant, decreasing slightly by 2%. This
it was recovering to the base case scenario of No shock rather than the means that the system resilience effectiveness during IS wasn’t

9
D. Al Assaad et al. Building and Environment 243 (2023) 110669

Fig. 6. Spider chart of resilience aspects (εres , εabs , εrec ) segregated as function of indoor contaminants for a MS for different doMS.

challenged as much as it was the case with MS. The degree of shock For DCV w/o filters for doIS ∈ [0.82, 1], the effect of filters on resil­
impact on exposure remained close to the systems’ base cases of No ience performance was not pronounced since εres,PM2.5 remained almost
shock. Note that the IS manifested for DCV later for doIS > 0.22. Since constant, decreasing slightly by 2%. εabs,PM2.5 increased from 0.054 to
there was still some deviation in εres,VOCs , it was still important to analyze 0.071 and εrec,PM2.5 decreased from 0.91 to 0.88 with increasing doIS.
the rest of the resilience aspects. εres,VOCs , εabs,VOCs & εrec,VOCs had also Looking at Fig. 8, across the entire doOS range for CAV, VOCs and
slight fluctuations: εabs,VOCs slightly increased linearly from 0.14 to 0.2 CO2 had similar resilience aspects (εres , εabs , εrec ) for doOS > 0.29. For
for CAV and from 0.07 to 0.09 for DCV. εrec,VOCs slightly decreased lin­ DCV w/o filters, for doOS > 0.58, CO2 and PM2.5 had also similar
early from 0.97 to 0.94 for CAV and from 0.52 to 0.49 for DCV. Similar resilience aspects.
trends were obtained for CO2 (Fig. S2). For CAV (Fig. 8, Fig. S3), with increasing doOS starting 0.58, εres,VOCs
εabs for CAV were higher than DCV, due to the lag time of the DCV in remained almost constant, decreasing slightly by 1%. This means that
maximizing the flow rate and thus, concentrations were peaking quickly the system resilience effectiveness during OS wasn’t challenged as much
to the maximum concentration before plateauing. As with MS, εrec,VOCs as it was the case with MS. It was even less than IS. Similar to εres,VOCs ,
for DCV were ∼ 5% higher than CAV. εrec,CO2 were ∼ 8.3% lower for εabs,VOCs & εrec,VOCs remained more or less constant at values of 0.38 &
DCV. This was since recovery time started for DCV from slightly higher 0.81 respectively. For CAV & DCV, εres,CO2 were similar and decreased
CO2 concentrations than CAV. In this specific case study, it can be due to linearly from 0.98 to 0.92. εabs,CO2 increased with increasing doOS from
the inherent hysteresis effect of the DCV PI control, causing the flow rate 0.34 to 0.38 and was similar for both CAV & DCV. This was since the OS
and thus concentrations to fluctuate slightly. began early in the morning 6:00 a.m. and thus DCV reacted earlier as

10
D. Al Assaad et al. Building and Environment 243 (2023) 110669

Fig. 7. Spider chart of resilience aspects (εres , εabs , εrec ) segregated as function of indoor contaminants for a IS for different doIS.

well. By the time occupants came in around 8:00 a.m., DCV and CAV had the HQ distribution (mean, 5th, 95th percentile) with increasing doS for
similar operation (i.e., maximum supply flow). εrec,CO2 also decreased each system and each HAP can be seen in Fig. 9. The resulting weights of
linearly with increasing doOS from 0.83 to 0.81 for DCV and from 0.85 each HAP can be seen in Fig. 10.
to 0.83 for CAV. εrec,CO2 was 2% lower for DCV as the latter was recov­ According to Fig. 9, for both CAV & DCV, and for MS, IS and OS,
ering from slightly higher concentrations generated by the hysteresis HQmean,VOCs (i.e., arithmetic mean of median concentrations in E120,
effect of the PI control. This effect was less pronounced than for IS. E220) was always higher >1 across the entire doS range. This meant that
For DCV w/o filters for doOS ∈ [0.58, 1], the effect of filters on VOCs’ concentrations (throughout the day) were always above the TRV.
resilience performance was more pronounced than IS, especially on the Thus, VOCs will always have a weight in the calculation of the RSHAPs
degree of impact and absorptivity of the shock, while it did not affect the component of the RS (Fig. 10).
recovery: εres,PM2.5 decreased linearly by 10%. εabs,PM2.5 increased line­ The HQmean,VOCs increased with increasing doMS linearly by 52.8%,
arly by 17.5%. εrec,PM2.5 remained almost constant at a value of 0.83. 47% for CAV & DCV respectively. For IS & OS, it remained at a constant
value of ~2.6 for CAV and ~3.6 for DCV (Fig. 9). This can be explained
4.2. Ranking of HAPs by analyzing the temporal distribution of VOC concentrations. In the
base case of No shock, for both CAV & DCV, the distribution of VOC
To calculate the final RS that aggregates all pollutants (Step 7 in concentrations was governed by build-up due to envelope emissions,
Fig. 1), it is first important to rank and weight the HAPs (Step 6, Fig. 1) which started occurring the moment the AHU turned off in the previous
namely VOCs, PM2.5 and PM10 using the combined quantitative & evening and turned on the next day morning [16]. In the case of MS, all
qualitative assessment method in Ref. [27] (Step 5, Fig. 1). The plots of values of VOC concentrations that were once below the base case

11
D. Al Assaad et al. Building and Environment 243 (2023) 110669

Fig. 8. Spider chart of resilience aspects (εres , εabs , εrec ) segregated as function of indoor contaminants for a OS for different doOS.

median value were surpassing it, shifting the HQmean,VOCs to higher 4.3. The total resilience score (RS)
values. This wasn’t the case for IS & OS, where the build-up peaks
remained the maximum. Fig. 11 illustrates the evolution of RS with increasing doMS, doIS &
The HQmean,PM2.5 remained below 1 for all systems and all shocks and doOS for CAV, DCV & DCV w/o filters (Step 7). A first glance at the RS
while it increased in the case of DCV w/o filters, it remained below 1. showed a common trend: For each type of shock and system, RS
However, the HQP95,PM2.5 did significantly cross 1 in the case of MS, for decreased quickly following a linear trend until a certain value of doS
doMS > 0.288, for both CAV & DCV and DCV w/o filters. The same trend where it either plateaued at a fixed final value or continued its decrease
was obtained for PM10. In this case only, both PM2.5 & PM10 had a at a slower pace. For the case of MS, RS of all three systems deteriorated
weight in the calculation of the RSHAPs as seen in Fig. 10a. In the case of quickly, decreasing by 64% by doMS = 0.288 (i.e., 2.8 h of power
DCV & DCV w/o filters, for doMS > 0.288, the weights of PM2.5 & PM10 outage). Beyond this value, RS decreased at a slower pace following a
surpassed that of VOCs (e.g., for DCV, doMS = 0.721: ωVOCs = 0.15, concave parabolic trend.
ωPM2.5 = 0.45, ωPM10 = 0.40) while they stayed below for CAV. Comparing the three systems, CAV & DCV ended up having the same
As HQP95,PM2.5 & HQP95,PM10 remained below 1 for IS & OS (Fig. 8), RS (Fig. 11a). Looking at Fig. 6 and Fig. S1(b), the resilience aspects of
for CAV (Fig. 10a) & DCV (Fig. 10b), only the VOCs influenced the CO2 for both systems were the same, resulting in equal RSCO2 . For
RSHAPs . In the case of DCV w/o filters for IS & OS, only PM2.5 had a RSHAPs , the values between systems evened out: They had the same
weight in the RSHAPs for doIS > 0.62 & doOS > 0.29 exceeding that of resilience aspects on PM with DCV having higher ωPM normally resulting
VOCs, notably in the case of OS. These results were also reflected in the in higher RSHAPs . However, CAV had a lower εrec,VOCs with higher ωVOCs
analysis of the εres in the previous section. evening out the RSHAPs between systems. For doMS > 0.288, DCV w/o

12
D. Al Assaad et al. Building and Environment 243 (2023) 110669

Fig. 9. HQ of the HAPs for a) CAV, b) DCV, and c) DCV w/o filters for MS, IS and OS with increasing doS.

filters had a 13% lower RS than the other systems given the effect of PM rates than CAV, followed by DCV w/o filters and finally by CAV due to
and their high ωPM . The final RS at doMS = 1, were 0.22 for CAV & DCV its constant operation at the maximum flow rate of 4400 m3/h.
and 0.19 for DCV w/o filters. Finally, comparing the RS of the three shock types per system
For the case of IS, RS of CAV deteriorated by 38.8% starting doIS = (Fig. S4), for all three systems, OS was the least critical shock. For CAV,
0.07 until doIS = 0.22 (i.e., 12 additional occupants). For DCV & DCV w/ this was followed by IS and finally MS. For DCV and DCV w/o filters, it
o filters, it deteriorated by 46% starting doIS = 0.07 until doIS = 0.42 (i. depended on the range of doS:
e., 24 additional occupants). Until doIS = 0.42, DCV had a better resil­
ience performance than CAV. This is since DCV did not react to the IS • For 0.15 < doS < 0.4, & 0.8 < doS < 1. IS was more critical than MS
until doIS > 0.22. Beyond this value until doIS = 1, RS plateaued at a • For 0.4 < doS < 0.8, MS was more critical than IS.
value of 0.43 for CAV and 0.36 for DCV. In this range, the CAV had the
better resilience performance due to overall higher values of the resil­ 5. Limitations and outlook
ience aspects (Fig. 7, Fig. S2). The effect of filters did not reflect in the
RS. This is expected for IS & MS since the buildup of indoor pollutants’ The present framework can holistically assess the IAQ resilience of
concentrations was governed by indoor sources unlike OS. any building typology equipped with any type of ventilation system(s)
For the case of OS, RS of CAV deteriorated by 51% until doOS = 0.58 and control strategy (i.e., mechanical, hybrid, natural, localized), with
(i.e., 1.5 × increase in outdoor pollution). For DCV and DCV w/o filters, knowledge of the building’s occupancy patterns, indoor/outdoor
it deteriorated by 26% until doOS = 0.29 (i.e., 72% increase in outdoor pollution sources. It can be used to design ventilation systems, their
pollution). Up until doOS = 0.29, all three systems had the same RS. control strategies, other IAQ management strategies (e.g., filtration,
Beyond doOS = 0.58, the RS of CAV plateaued at 0.49. Beyond doOS = carbon capture techniques [66,67]) & influencing building parameters
0.29, the RS of DCV plateaued at 0.73. However, that of DCV w/o filters (e.g., envelope tightness). This can be done while considering cost
kept decreasing at a slower pace reaching a final value of 0.49 at doOS = constraints (i.e., energy, installation, maintenance). It can be used to test
1 like that of CAV. For an OS, DCV had the best RS due to lower flow the effect of occupant behavior and interaction with the building.

13
D. Al Assaad et al. Building and Environment 243 (2023) 110669

Fig. 10. Ranking of HAPs for CAV, DCV, and DCV w/o filters for a) MS, b) IS and c) OS with increasing doS.

Moreover, by considering the effect of air distribution and 6. Conclusion


non-homogenous IAQ fields (e.g., through computational fluid dy­
namics simulations), other ventilation design parameters could be In this work, a holistic quantitative assessment framework of IAQ
investigated (e.g., positioning of inlet & outlet, grill design). resilience was developed and applied in a case study educational
Another advantage of this framework is its multi-layer structure building using a previously validated model. The resulting resilience
allowing to assess IAQ resilience at the individual contaminant level score (RS) was calculated for different shock types (MS, IS & OS) and
while visualizing & analyzing the different resilience aspects separately different systems: CAV, DCV and DCV w/o filters. The key takeaways
using the radar chart visualization. This framework can also be extrap­ from applying this framework were summarized below:
olated to any type of resilience discipline, given that the corresponding
aspects of resilience are correctly identified. An example of this is • MS challenged the resilience of all systems against all contaminants
thermal resilience to over- or under-heating events in the context of (CO2, VOCs, PM2.5, PM10). IS challenged the resilience of CO2 and
climate change, where holistic indicators and quantitative assessment VOCs for CAV and DCV and additionally PM2.5 for DCV w/o filters
approaches are still missing [68]. This will be part of future work. but only for extreme IS (doIS > 0.62). OS challenged the resilience of
The current indicator is a variable function of doS. While it allows to CO2 and VOCs for CAV and DCV and additionally PM2.5 for DCV w/o
compare and rank systems, it does not allow to label them into cate­ filters for doOS > 0.29.
gories. To achieve this, further work is needed on the indicator through • εres , εabs , εrec trends depend on doS, the building’s occupancy patterns
an additional layer of calculation and simulations. This will also be a and the speed of absorption of contaminants, especially for MS. For
follow-up of the current study. Another follow-up study includes IS and OS, the rate of change of εres , εrec , εres was linear and slow
applying the framework to consider different shock combinations. Such across the doS range.
a study could give insight into the weight of concurrent shocks on RS • For MS, filters did not affect εabs nor εrec , only εres which was ∼ 10%
and the calculation of a total doS. lower for DCV w/o filters. For IS, the effect of filters on resilience
performance was not that pronounced since εres,PM2.5 remained
almost constant. For OS, the effect of filters on resilience

14
D. Al Assaad et al. Building and Environment 243 (2023) 110669

Fig. 11. Evolution of the total RS for CAV, DCV, and DCV with no filters for a) MS, b) IS and c) OS with increasing doS.

performance was more pronounced, especially on the degree of • For MS, CAV & DCV ended up having same RS and DCV w/o filters
impact and absorptivity of the shock, while it did not affect the re­ 13% lower RS. For IS, until doIS = 0.42, DCV had a better resilience
covery. Thus, in the case where a G4/F7 filters’ combination – typical performance than CAV. Beyond this value until doIS = 1, CAV had
of Passivhaus buildings, the trend of RS (Fig. 11) would start to shift the better resilience performance due to overall higher values of
from the trend followed by DCV to that of DCV w/o filters with resilience aspects. The effects of filters did not influence the RS for
decreasing filter performance. It would be most noticeable in the IS. For an OS, DCV had the best RS, followed by DCV w/o filters and
case of OS. finally by CAV.
• For each type of shock and system, RS decreased quickly following a • The most critical shock type depends on the system type and doS. For
linear trend until a certain value of doS where it either plateaued at a CAV & DCV, OS was the least critical. For CAV, this was followed by
fixed final value or continued its decrease at a slower pace. IS and MS. For DCV and DCV w/o filters, for 0.15 < doS < 0.4, & 0.8

15
D. Al Assaad et al. Building and Environment 243 (2023) 110669

< doS < 1. IS was more critical than MS and for 0.4 < doS < 0.8, MS [12] C. Hughes, S. Natarajan, Summer thermal comfort and overheating in the elderly,
Build. Serv. Eng. Res. Tecnol. 40 (2019) 426–445, https://doi.org/10.1177/
was more critical than IS.
0143624419844518.
[13] S.R. Thom, V.M. Bhopale, J.P. Hu, M. Yang, Increased carbon dioxide levels
This conclusion is a result of aggregating the different resilience as­ stimulate neutrophils to produce microparticles and activate the nucleotide-
pects and contaminants. This rounds the conclusion to the authors’ binding domain-like receptor 3 inflammasome, Free Radic. Biol. Med. 106 (2017)
406–416, https://doi.org/10.1016/J.FREERADBIOMED.2017.03.005.
previous work [16] that compared shock types only based on ppm.hours [14] H. Guo, S.C. Lee, L.Y. Chan, W.M. Li, Risk assessment of exposure to volatile
per contaminant. organic compounds in different indoor environments, Environ. Res. 94 (2004)
57–66, https://doi.org/10.1016/S0013-9351(03)00035-5.
[15] C. Zhang, O.B. Kazanci, R. Levinson, P. Heiselberg, B.W. Olesen, G. Chiesa,
CRediT authorship contribution statement B. Sodagar, Z. Ai, S. Selkowitz, M. Zinzi, A. Mahdavi, H. Teufl, M. Kolokotroni,
A. Salvati, E. Bozonnet, F. Chtioui, P. Salagnac, R. Rahif, S. Attia, V. Lemort,
E. Elnagar, H. Breesch, A. Sengupta, L.L. Wang, D. Qi, P. Stern, N. Yoon, D.
Douaa Al Assaad: Writing – review & editing, Writing – original I. Bogatu, R.F. Rupp, T. Arghand, S. Javed, J. Akander, A. Hayati, M. Cehlin,
draft, Visualization, Validation, Software, Methodology, Formal anal­ S. Sayadi, S. Forghani, H. Zhang, E. Arens, G. Zhang, Resilient cooling strategies – a
ysis, Data curation, Conceptualization. Abantika Sengupta: Validation, critical review and qualitative assessment, Energy Build. 251 (2021), 111312,
https://doi.org/10.1016/J.ENBUILD.2021.111312.
Software, Data curation. Hilde Breesch: Writing – review & editing,
[16] D. Al Assaad, A. Sengupta, H. Breesch, Demand-controlled ventilation in
Supervision, Resources, Project administration, Investigation, Funding educational buildings: energy efficient but is it resilient? Build. Environ. 226
acquisition. (2022), 109778 https://doi.org/10.1016/j.buildenv.2022.109778.
[17] T.J. Pettit, J. Fiksel, K.L. Croxton, Ensuring supply chain resilience: development of
a conceptual framework, J. Bus. Logist. 31 (2010) 1–21, https://doi.org/10.1002/
J.2158-1592.2010.TB00125.X.
Declaration of competing interest [18] S. Hosseini, K. Barker, J.E. Ramirez-Marquez, A review of definitions and measures
of system resilience, Reliab. Eng. Syst. Saf. 145 (2016) 47–61, https://doi.org/
10.1016/J.RESS.2015.08.006.
The authors declare that they have no known competing financial
[19] A. Rose, Economic resilience to natural and man-made disasters: multidisciplinary
interests or personal relationships that could have appeared to influence origins and contextual dimensions, Environ. Hazards 7 (2007) 383–398, https://
the work reported in this paper. doi.org/10.1016/J.ENVHAZ.2007.10.001.
[20] L. Ji, C. Shu, A. Laouadi, M. Lacasse, L.L. Wang, Quantifying Building and Zone
Level Thermal Resilience against Summertime Heat Events, Available at: SSRN
Data availability 4210569. (n.d.).
[21] S. Homaei, M. Hamdy, Thermal resilient buildings: how to be quantified? A novel
Data will be made available on request. benchmarking framework and labelling metric, Build. Environ. 201 (2021),
108022, https://doi.org/10.1016/J.BUILDENV.2021.108022.
[22] A. Sengupta, D. Al Assaad, J.B. Bastero, M. Steeman, H. Breesch, Impact of
Acknowledgements heatwaves and system shocks on a nearly zero energy educational building: is it
resilient to overheating? Build. Environ. 234 (2023), 110152 https://doi.org/
10.1016/j.buildenv.2023.110152.
This work has been supported by the Flemish Agency for Innovation [23] E. Tavakoli, A. O’Donovan, M. Kolokotroni, P.D. O’Sullivan, Evaluating the indoor
and Entrepreneurship (VLAIO) in the Flux50 project ‘Towards Smart thermal resilience of ventilative cooling in non-residential low energy buildings: a
review, Build. Environ. 222 (2022), 109376, https://doi.org/10.1016/J.
Ventilation’ (HBC.2020.2520). BUILDENV.2022.109376.
[24] M. Hendel, K. Azos-Diaz, B. Tremeac, Behavioral adaptation to heat-related health
Appendix A. Supplementary data risks in cities, Energy Build. 152 (2017) 823–829, https://doi.org/10.1016/J.
ENBUILD.2016.11.063.
[25] R. Rahif, M. Hamdy, S. Homaei, C. Zhang, P. Holzer, S. Attia, Simulation-based
Supplementary data to this article can be found online at https://doi. framework to evaluate resistivity of cooling strategies in buildings against
org/10.1016/j.buildenv.2023.110669. overheating impact of climate change, Build. Environ. 208 (2022), 108599,
https://doi.org/10.1016/J.BUILDENV.2021.108599.
[26] M. Hamdy, S. Carlucci, P.J. Hoes, J.L.M. Hensen, The impact of climate change on
References the overheating risk in dwellings—a Dutch case study, Build. Environ. 122 (2017)
307–323, https://doi.org/10.1016/J.BUILDENV.2017.06.031.
[27] G. Sérafin, P. Blondeau, C. Mandin, Indoor air pollutant health prioritization in
[1] T.J. Vogus, K.M. Sutcliffe, Organizational resilience: towards a theory and research
office buildings, Indoor Air 31 (2021) 646–659, https://doi.org/10.1111/
agenda, Conf Proc IEEE Int Conf Syst Man Cybern (2007) 3418–3422, https://doi.
INA.12776.
org/10.1109/ICSMC.2007.4414160.
[28] Y. Zhao, J. Wen, F. Xiao, X. Yang, S. Wang, Diagnostic Bayesian networks for
[2] A. Rose, S.Y. Liao, Modeling regional economic resilience to disasters: a
diagnosing air handling units faults – part I: faults in dampers, fans, filters and
computable general equilibrium analysis of water service disruptions, J. Reg. Sci.
sensors, Appl. Therm. Eng. 111 (2017) 1272–1286, https://doi.org/10.1016/J.
45 (2005) 75–112, https://doi.org/10.1111/J.0022-4146.2005.00365.X.
APPLTHERMALENG.2015.09.121.
[3] M.D. Turner, Political ecology I: an alliance with resilience? Prog. Hum. Geogr. 38
[29] Power Cuts in Belgium, .https://Www.Fluvius.Be/Nl/Meer-Weten/Stroomonde
(2013) 616–623, https://doi.org/10.1177/0309132513502770.
rbrekingen?App-Refresh=1638796478670 (n.d.).
[4] B.D. Youn, C. Hu, P. Wang, Resilience-driven system design of complex engineered
[30] C.J. Weschler, Roles of the human occupant in indoor chemistry, Indoor Air 26
systems, J. Mech. Design, Trans. ASME 133 (2011), https://doi.org/10.1115/
(2016) 6–24, https://doi.org/10.1111/ina.12185.
1.4004981/467326.
[31] G. Lissens, W. Debruyn, C. Mensink, G. Dumont, SMOGSTOP: a model for
[5] The 2003 Northeast Blackout–Five Years Later - Scientific American, (n.d.). https
forecasting maximum daily ozone concentration in Belgium, Environmetrics 11
://www.scientificamerican.com/article/2003-blackout-five-years-later/(accessed
(2000) 511–521, https://doi.org/10.1002/1099–095X(200009/10)11:5<511::
March 27, 2023).
AID-ENV415>3.0.CO;2-G.
[6] J. Manuel, The long road to recovery: environmental health impacts of hurricane
[32] H.T. Tran, M. Balchanos, J.C. Domerçant, D.N. Mavris, A framework for the
sandy, Environ. Health Perspect. 121 (2013), https://doi.org/10.1289/EHP.121-
quantitative assessment of performance-based system resilience, Reliab. Eng. Syst.
A152.
Saf. 158 (2017) 73–84, https://doi.org/10.1016/J.RESS.2016.10.014.
[7] W. Mao, V.I.O. Agyapong, The role of social determinants in mental health and
[33] D. Norbäck, K. Nordström, Sick building syndrome in relation to air exchange rate,
resilience after disasters: implications for public health policy and practice, Front.
CO2, room temperature and relative air humidity in university computer
Public Health 9 (2021) 551, https://doi.org/10.3389/FPUBH.2021.658528/
classrooms: an experimental study, Int. Arch. Occup. Environ. Health 82 (2008)
BIBTEX.
21–30, https://doi.org/10.1007/S00420-008-0301-9/TABLES/6.
[8] K. Riahi, S. Rao, V. Krey, C. Cho, V. Chirkov, G. Fischer, G. Kindermann,
[34] M. Hodgson, Sick building syndrome., Occup. Med. 15 (2000) 571–585. http
N. Nakicenovic, P. Rafaj, RCP 8.5-A scenario of comparatively high greenhouse gas
s://europepmc.org/article/med/10903551 (accessed November 11, 2021).
emissions, Clim. Change 109 (2011) 33–57, https://doi.org/10.1007/S10584-011-
[35] X. Zhang, P. Wargocki, Z. Lian, Effects of exposure to carbon dioxide and human
0149-Y/FIGURES/12.
bioeffluents on cognitive performance, Procedia Eng. 121 (2015) 138–142,
[9] C.S. Holling, Resilience and stability of ecological systems, Annu. Rev. Ecol.
https://doi.org/10.1016/J.PROENG.2015.08.1040.
Systemat. 4 (1973) 1–23, https://doi.org/10.1146/annurev.es.04.110173.000245.
[36] REHVA Journal CO₂ monitoring and indoor air quality (n.d.), https://www.rehva.
[10] R. Francis, B. Bekera, A metric and frameworks for resilience analysis of engineered
eu/rehva-journal/chapter/co2-monitoring-and-indoor-air-quality, 2023. June 22.
and infrastructure systems, Reliab. Eng. Syst. Saf. 121 (2014) 90–103, https://doi.
[37] L. Rueda, K. Agbossou, A. Cardenas, N. Henao, S. Kelouwani, A comprehensive
org/10.1016/J.RESS.2013.07.004.
review of approaches to building occupancy detection, Build. Environ. 180 (2020),
[11] M. Ungar, Qualitative contributions to resilience research, Qual. Soc. Work 2
106966, https://doi.org/10.1016/J.BUILDENV.2020.106966.
(2003) 85–102, https://doi.org/10.1177/1473325003002001123.

16
D. Al Assaad et al. Building and Environment 243 (2023) 110669

[38] S.J. Emmerich, A.K. Persily, Literature review on CO2-based demand-controlled in a nZEB educational building by monitoring and simulation, Energy Build. 231
ventilation, Trans. Am. Soc. Heat. Refrigerat. Air Conditioning Eng. 103 (1997) (2021), 110616, https://doi.org/10.1016/J.ENBUILD.2020.110616.
229–243. [54] B. Merema, M. Delwati, M. Sourbron, H. Breesch, Demand controlled ventilation
[39] J.M. Logue, T.E. Mckone, M.H. Sherman, B.C. Singer, Hazard assessment of (DCV) in school and office buildings: lessons learnt from case studies, Energy Build.
chemical air contaminants measured in residences, Indoor Air 21 (2011) 92–109, 172 (2018) 349–360, https://doi.org/10.1016/J.ENBUILD.2018.04.065.
https://doi.org/10.1111/J.1600-0668.2010.00683.X. [55] A. Persily, L. de Jonge, Carbon dioxide generation rates for building occupants,
[40] K. Azuma, I. Uchiyama, S. Uchiyama, N. Kunugita, Assessment of inhalation Indoor Air 27 (2017) 868–879, https://doi.org/10.1111/INA.12383.
exposure to indoor air pollutants: screening for health risks of multiple pollutants [56] indicators.be - CO2 atmospheric concentration, n.d.), https://indicators.be/en/i/
in Japanese dwellings, Environ. Res. 145 (2016) 39–49, https://doi.org/10.1016/ G13_CO2/CO2_atmospheric_concentration, 2022. March 2.
J.ENVRES.2015.11.015. [57] Hoge Gezondheidsraad, Indoor Air Quality in Belgium, HGR, Brussels, 2017.
[41] R.D.E.F. Bromes, Hierarchisation sanitaire des parametres D’INTERET pour [58] X. Tang, P.K. Misztal, W.W. Nazaroff, A.H. Goldstein, Volatile organic compound
L’OBSERVATOIRE de la QUALITE de L’AIR INTERIEUR, in: APPLICATION AUX emissions from humans indoors, Environ. Sci. Technol. 50 (2016) 12686–12694,
PHTALATES, PARAFFINES CHLOREES A CHAINE COURTE, ORGANO-ETAINS, https://doi.org/10.1021/acs.est.6b04415.
ALKYL PHENOLS ET, 2005. [59] D.H. Do, H. van Langenhove, C. Walgraeve, S.F. Hayleeyesus, P. de Wispelaere,
[42] Chemical Search | IRIS | US EPA, (n.d.). https://cfpub.epa.gov/ncea/iris/search/ J. Dewulf, K. Demeestere, Volatile organic compounds in an urban environment: a
index.cfm (accessed April 3, 2023). comparison among Belgium, Vietnam and Ethiopia, Int. J. Environ. Anal. Chem. 93
[43] W.H. Organization, WHO guidelines for indoor air quality: selected pollutants, in: (2013) 298–314.
World Health Organization, Regional Office for Europe, 2010. [60] D. Al Assaad, K. Ghali, N. Ghaddar, C. Habchi, Coupled CFD and particle
[44] Indoor Environment Decree, Decree of the Flemish Government of 11 June 2004 resuspension models under combined effect of mechanical and aerodynamic
concerning measures to combat health risks due to pollution of the indoor disturbances, Build. Environ. 169 (2020), 106567, https://doi.org/10.1016/J.
environment, n.d. Care and Health (2021) https://www.zorg-en-gezondheid.be/ BUILDENV.2019.106567.
binnenmilieubesluit-besluit-van-de-vlaamse-regering-van-11-juni-2004-houdende- [61] A.R. Ferro, R.J. Kopperud, L.M. Hildemann, Source strengths for indoor human
maatregelen-tot. December 3 activities that resuspend particulate matter, Environ. Sci. Technol. 38 (2004)
[45] EU-LCI values (n.d.), https://single-market-economy.ec.europa.eu/sectors/constr 1759–1764, https://doi.org/10.1021/ES0263893/SUPPL_FILE/
uction/eu-lci-subgroup/eu-lci-values_en. (Accessed 3 April 2023). ES0263893SI20030924_074503.PDF.
[46] A.J. Baars, R.M.C. Theelen, P.J.C.M. Janssen, M.C.M. Meijerink, L. Verdam, M. [62] Baudelopark air quality index (AQI) and gent air pollution, IQAir (2023) (n.d.),
J. Zeilmaker, J.M. Hesse, M.E. Van Apeldoorn, Re-evaluation of Human- https://www.iqair.com/belgium/flanders/gent/baudelopark. May 12.
Toxicological Maxi-Mum Permissible Risk Levels, 2001. https://www.rivm. [63] C.A. Faulkner, J.E. Castellini, W. Zuo, D.M. Lorenzetti, M.D. Sohn, Investigation of
nl/bibliotheek/rapporten/711701025.pdf. (Accessed 3 April 2023). HVAC operation strategies for office buildings during COVID-19 pandemic, Build.
[47] Indoor air quality resources for professionals, Canada.ca (2023) n.d. https://www. Environ. 207 (2022), 108519, https://doi.org/10.1016/J.
canada.ca/en/health-canada/services/air-quality/residential-indoor-air-quali BUILDENV.2021.108519.
ty-guidelines.html. April 3 [64] L. Wallace, Indoor particles: a review, J. Air Waste Manag. Assoc. 46 (1996)
[48] Menerga | Menerga, (n.d.). https://www.menerga.com/(accessed February 3, 98–126, https://doi.org/10.1080/10473289.1996.10467451.
2022). [65] W.H. Organization, WHO Global Air Quality Guidelines: Particulate Matter (PM2.
[49] J. van Herreweghe, S. Caillou, T. Haerinck, F. Graindorge, C. Delmez, P. van 5 and PM10), Ozone, Nitrogen Dioxide, Sulfur Dioxide and Carbon Monoxide:
Huffel, Real-life ventilation filter performance: final results of an in-depth study, in: Executive Summary, 2021.
2nd AIVC Conference 10th TightVent Conference 8th Venticool Conference, 2022. [66] J.P. Harrouz, J. Karam, K. Ghali, N. Ghaddar, Personalized ventilation with
Rotterdam, Netherlands. embedded air treatment system for simultaneous cooling and sorption-based
[50] Dymola - Dassault Systèmes® (n.d.), https://www.3ds.com/products-services/cat carbon and humidity capture, Energy Convers. Manag. 291 (2023), 117290,
ia/products/dymola/. (Accessed 11 February 2022). https://doi.org/10.1016/J.ENCONMAN.2023.117290.
[51] F. Jorissen, G. Reynders, R. Baetens, D. Picard, D. Saelens, L. Helsen, [67] J.P. Harrouz, K. Ghali, M. Hmadeh, N. Ghaddar, S. Ghani, Feasibility of MOF-based
Implementation and verification of the IDEAS building energy simulation library, carbon capture from indoor spaces as air revitalization system, Energy Build. 255
J. Build. Perform Simul. 11 (2018) 669–688, https://doi.org/10.1080/ (2022), 111666, https://doi.org/10.1016/J.ENBUILD.2021.111666.
19401493.2018.1428361. [68] A. Sengupta, H. Breesch, D. Al Assaad, M. Steeman, Evaluation of Thermal
[52] B. Merema, H. Breesch, D. Saelens, MPC Framework for All-Air Systems in Non- Resilience to Overheating for an Educational Building in Future Heatwave
residential Buildings, KU Leuven, 2021. Scenarios, 2023, https://doi.org/10.1080/14733315.2023.2218424. Https://Doi.
[53] M. Borrelli, B. Merema, F. Ascione, R. Francesca De Masi, G. Peter Vanoli, Org/10.1080/14733315.2023.2218424.
H. Breesch, Evaluation and optimization of the performance of the heating system

17

You might also like