IoT SENSING FOR BUILDING RESILIENCE ... › sites › default › files › sbb...Nishant Kumar, Krishnanand Kaippilly Radhakrishnan, Sanjib Kumar Panda The significance of the economic

Fall 2019Volume 1

IoT SENSING FOR BUILDING RESILIENCE CHARACTERIZATION

02

TECHNOLOGY NEWS 13

GLARE: BAYESIAN ANALYSIS METHOD | CYBER SECURITY FOR SMART BUILDING IOT SYSTEM | MULTI-CRITERIA DECISION-MAKING FOR BUILDING DESIGN | WEARABLE SENSORS TO PREDICT THERMAL COMFORT| SINBERBEST DAYLIGHT EMULATOR

05

In November of 2018, North California was highlyaffected for weeks by wildfire in Chico. Air pollutionspread thorough the San Francisco Bay Area. As aresult, many businesses and schools including Universityof California Berkeley campus were closed. Air pollutioncaused disruption to many but not all businesses andschools. Variety of responses across Bay Area showedlack of policies on how to respond to the extreme airpollution event and revealed high level of confusion andchaos.

Wildfire smoke is a toxic mixture of particulate matter(PM), carbon dioxide, carbon monoxide, complexhydrocarbons, nitrogen oxides, and many othercompounds. Exposure to wildfire smoke is associatedwith respiratory problems including asthma, chronicobstructive pulmonary disease, bronchitis andpneumonia. Studies have also shown that exposure towildfire during pregnancy may impact birth weightamong term infants, and increases in mental healthsymptoms among adolescents.

While the health risks are clear, standards such ASHRAE62.1 and 62.2 only address normal outdoor airconditions. A few guidelines exist for protecting buildingoccupants from pollution caused by wildfire: ‘WildfireSmoke Response Planning,’ by the British ColumbiaCenter for Disease Control; and ‘Wildfire SmokeGuidelines,’ from the California Department of PublicHealth. However, these provide very generalrecommendations such as “reduce infiltration” orinstalling “effective filters.” Such qualitative suggestionsare not sufficient to quantify building properties anddetermine how indoor conditions affect occupants.

In March 2018, the research team installed low‐cost IoTPM2.5 sensors on the roofs of two UC Berkeley buildingsand 14 indoor sensors in each building. During the 2018fires, these sensors enabled us to understand andevaluate building resilience to urban scale air pollution,and quantifying particle penetration. The buildingsincluded one mixed‐mode ventilated and onemechanically ventilated, both located in Berkeley, CA.Although the study was performed in the San FranciscoBay Area in California, wildfire smoke in Californiarepresents the same environmental problem Singaporefaces during human‐induced forest fires in Indonesia.Tools and approaches depicted in this study can bedirectly implemented in Singapore to characterizebuilding resilience during the haze period.

We also used this opportunity to evaluate an Internet ofThings (IoT) environmental sensing network, with sensorsinside and outside of buildings, combined with occupantsurveys.

From November 8th through November 21st, the twobuildings described below were impacted by wildfiresmoke:

Mechanically Ventilated Building (4thSt): This is amechanically ventilated building with a centralized HVACsystem. The HVAC system utilizes three stages of airfiltration: (1) a Minimum Efficiency Reporting Value(MERV) 8 pleated filter; (2) gas phase filter; and (3) ahigh‐efficiency MERV 13 filter. Measurements of CO2

decay showed that building is airtight, indicating thatinfiltration plays a minor role in bringing outdoorpollution indoors.

Mixed‐mode Ventilated Building (wurster): This is amixed‐mode operated building, with the east wingnaturally ventilated with a high level of infiltration.During normal operation, the building measurementsshowed that CO2 levels most of the time are below 600ppm.

Sensors usedClarity and Senseware PM2.5 nodes use the principle oflight scattering to count the particle number. Theaccuracy of the all the sensors was the same—within ±10μg/m3 in the range of 0 to 100 μg/m3 and ±10% in therange of 100 to 1000 μg/m3. Data was collected on 15‐minute intervals for Clarity nodes on 1‐minute intervalsfor Senseware nodes.

Indoor air quality during an air pollution episodeUsing an IoT sensing network, position outdoors andindoors, enabled us to understand the building operationin the context of indoor and outdoor PM2.5 levels. Resultsin Figure 1 show that the median hourly indoor PM2.5

concentration over the entire wildfire event was 21μg/m3 for the building on 4thSt building, and 36 μg/m3

for Wurster Hall. Measured CO2 decay rates, indicatinginfiltration of 0.15 air changes per hour (ACH), suggestthat 4thSt is airtight, showing that the ingress of PM2.5

through mechanical ventilation was the most dominant

COVERSTORY

2 - SinBerBEST sinberbest.berkeley.edu

IoT Sensing for Building Resilience CharacterizationJovan Pantelic and Megan Dawe

FIGURE 1 Outdoor (L) and Indoor (R) PM2.5 monitors

sinberbest.berkeley.edu SinBerBEST - 3

mechanism of PM2.5 penetration into the building.Combination of PM2.5 and CO2 sensing shows the keybuildings aspect where intervention can have the highestimpact.

Wurster Hall has operable windows designed to allow thenatural flow of air through the building, and althoughmonitored windows were closed (based on contactsensor readings) during the pollution event Wurster Hallhad infiltration of 0.40 ACH.

Our measurements enabled us to calculate an indoor tooutdoor (I/O) particle ratio as one of the keyperformance measures of the building. The I/O ratio willtell us how much outdoor pollution penetrated into thebuilding. For the entire period, the I/O ratio for the 4thStbuilding was 0.27, while for Wurster Hall I/O was 0.67.

Results from Figure 2 and calculated I/O representoverall operation of the building during the entirepollution episode. Combination of PM2.5 and CO2 sensingshows the key buildings aspect where intervention canhave the highest impact. Results in Figure 2 do not tell usmuch about the day‐to‐day operation, nor will it tell ushow these levels of pollution impacted buildingoccupants. Results as shown in Figure 2 tell us that the4thSt building is more resilient to extreme pollutionevents than Wurster Hall.

Occupant exposure to pollutantsPM2.5 concentration and length of the exposure periodaffect health, productivity and comfort of buildingoccupants. In the context of building resilience, wesimultaneously evaluated ambient concentration andexposure time. We compared the hourly median PM2.5

concentration to World Health Organization (WHO) 24‐hour exposure guidelines and calculated the exceedanceindex (E‐index). Results in Figure 2 also show ranking ofE‐index in the context of extreme pollution eventtimeline.

Results in Figure 3 show that 4thSt E‐index was below theWHO suggested threshold 66% of the time, compared toonly 23% of the time for Wurster Hall. We can calculateoverall E‐index by summing up hourly values for theentire pollution episode. The overall E‐index for 4thSt is0.82, suggesting that the building was resilient to airpollution and that filtration with the MERV 8 and MERV13 filters was able to reduce the PM2.5 penetration tomaintain normal operation inside the building. ForWurster Hall, the average E‐index was 1.69, suggestingthat air quality conditions could potentially have severeconsequences on those with respirtory health issues.

FIGURE 2 Comparison of hourly PM2.5 concentrations between the two sites for the entire period of air pollution episode. The box plot represents the 25th percentile, median, and 75th percentile, and the whiskers represent the 5th and 95th percentiles.

COVERSTORY


Conclusion and futureThe tool we developed can be used to quantify in realtime building resiliency to extreme air pollution episodes(wildfire generated smoke). Such analytical tool mightalso offer the ability to provide information for planningand in real‐time to identify buildings that can serve asclean‐air shelters during future fire events or point outmost effective interventions. Future studies should bescaled up to a larger cohort of buildings. Unfortunately,such fire events are expected to be more likely due toclimate change and continued development in forestedareas.

ReferencesPantelic J, Dawe M, Licina D (2019) Use of IoT sensing andoccupant surveys for determining the resilience ofbuildings to forest fire generated PM2.5. PLoS ONE 14(10):e0223136.https://doi.org/10.1371/journal.pone.0223136

FIGURE 3 Exceedance index PM2.5 heat map calculated for WHO 24 hour exposure guidelines at 4th Street (top) and Wurster Hall (bottom)


Being able to replicate findings is a fundamentalpart of conducting reliable research. Whenattempting to replicate glare studies, results haveled to inconsistent findings. Even when using thesame glare prediction model, studies have seldomused statistical tests designed to compare resultswith each other, and data used in these studieshave not commonly been publicly shared. We thinkthese practices must change to make buildingscience research more reliable, efficient in the useof resources and effective. A similar improvement ishappening in the field of thermal comfort, forexample, with the development of a databasemerging the results of thermal comfort studies donein real buildings all over the globe (database,paper).

To address this replicability issue, we proposed aBayesian approach that allows us to build on the work ofothers. To test it, we applied it to a specific problem inglare research — the effect of order bias that wedetected in our previous work (order bias occurs whensubjective criteria are evaluated in a strict sequence, forexample, first the lowest criterion and then others inincreasing order of magnitude). We performed alaboratory test with 55 participants and asked them toevaluate glare using a newly developed glare scale

created by the Berkeley Education Alliance for Researchin Singapore (‘BEARS’), shown in Figure 2.

HUMAN BUILDING NEXUS

Bayesian Analysis Method to Improve Research on GlareMichael Kent, Toby Cheung, Sergio Altomonte, Aleksandra Lipczyńska and Stefano Schiavon

FIGURE 1 Example of a high dynamic image constructed from the seven low dynamic images captured using a camera with fish‐eye‐lens (left); False‐colour Photosphere luminance map with Radiance image formatting (centre); Evalglare image with a blue circle at the centre of the screen representing the point of visual fixation. Pixels highlighted in pink surrounding the screen represent the identified glare sources (right).

FIGURE 2 The BEARS scale of subjective glare evaluation used to measure the magnitude of visual discomfort.


We performed the tests in three order sequences:ascending, descending, and randomized. We foundthat the study participants have different thresholdsdepending on the order of their exposure, as shown inthe figure 3. We used Bayesian inferential tests tocompare the effect of order bias detected in theprevious study to results of the new experiment. Theresults showed a close replication, highlighting thatthe order bias effect found in the original study wasalso present in the new data.

We believe that the wider application of Bayesianmethods and the making of the collected data open‐source can lead to improved and new glare models.This in turn, will help provide better design solutions.

Today, to increase the user comfort level, allcommercial and residential buildings areintegrated with smart electronics appliances,which provide user‐friendly high‐level operationand wide range of control ability. The wide rangeof control ability is possible due to the involvement ofthe Internet of Things (IoT), which provides seamlessinteractions between physical environment and the

Reference

Kent MG, Cheung T, Altomonte S, Schiavon S and Lipczyńska A. 2018. A Bayesian method of evaluating discomfort due to glare: The effect of order bias from a large glare source. Building and Environment 146: 258‐267. https://doi.org/10.1016/j.buildenv.2018.10.005(open‐access)

smart devices. The IoT ecosystem is formed byenabling sensors, actuators, data canters, middlewareand communication network, where differenttechnologies are involved, such as networkingtechnology, software technology, control technologyetc., for information and optimal decision making.

FIGURE 3 Mean Daylight Glare Index values for the three sequences and the four glare criteria found on the BEARS scale. The error bars show the standard deviations.

SMART & ENERGY TECHNOLOGIES

Assessment of Cyber Security for Smart Building IoT SystemNishant Kumar, Krishnanand Kaippilly Radhakrishnan, Sanjib Kumar Panda

The significance of the economic impact of IoT enabledsmart buildings is well proven on the commercial level aswell as at governmental level. It provides an optimal usestrategy for appliances, without compromising usercomfort and minimum electricity charges, which reducesthe peak demand on the grid and provides financialsupport to the consumer. It also optimises the use ofparking space, which is the key issue of every city. Videosurveillance and its analysis become very much easier,which is manually impossible. Due to real‐timedistributed information, water and electricity demandforecasting, planning and quality management becomeeasier, as well as during fault, it helps in locationdetection. Therefore, today data can be considered asblood and IoT system is the heart of the society.

However, high flexibility always comes with a high‐security issue. Similarly, in IoT enabled smart building,cybersecurity is the key challenge. Previously, differentmanufacturers of the different buildings follow differentautomation and communication protocols, whichprovides space to add compatible data addition ormanipulation.

In the modern automation and communication system ofSmart Building Management (SBM) uses advanced andcommon protocols such as Operational Technology (OT)protocol follows ModBus or BACnet and InformationTechnology (IT) protocol follows HTTP or FTP. However,still, the junction node of OT and IT is the key point forhackers. Easy disconnection and coupling on the junctionnode give a route for traffic analysis. Through this trafficreconnaissance, hacker performs scanning of IP addressesand vulnerabilities of the system. After that the hackercan go for man‐in‐the‐middle attacks, channels jamming,plantation of viruses, or can take permanent access asadministrator. Through these they can easily retract theimportant data, which harms individual or damage on asocietal level. Therefore the “ De Facto OperationalModel” for SBM needs to change.

Setting aside organisational barriers and acknowledgingthe IT/OT disconnect is the critical first step towards

implementing and operating cyber‐secure smart buildingcontrol systems. Therefore, try to follow IEC 62443 globalset of cybersecurity standards, which improve safety,availability, integrity and confidentiality of systems usedfor personal/commercial/industrial automation andcontrol. From a user point of view, the five key steps forsecure SBM system are as follows:

1. Assess and protect legacy OT building control systems.2. Choose IoT devices and vendors that follow a securedevelopment lifecycle approach.3. Implement secure OT building control systemarchitectures.4. Bridge the secure OT building control systems throughan IT security monitoring zone.5. Keep all software UpToDate and run full scanperiodically.

From a developer or support provider point of view, thefive key steps for secure SBM system are as follows:

1. Teams of OT and IT must work together to create amore secure system.2. Always perform testing on recent practical data.3. Frequently provide updates for every new bug.4. Divide the limit of overall control access in three parts,which are user access limit, customer support provideraccess limit and developer core committee access limit.5. Provide freeze option and factory reset option forcritical conditions.

Apart from these, the role of IMDA (Info‐communicationsMedia Development Authority) and theirIoT/cybersecurity initiatives are also very important,which needs to make policies and standard for types ofequipment. These policies secure data security and,standards secure equipment quality, which are necessaryfor the open market for consumer security. In the end,the most important role in cybersecurity is of userintelligence because majority of the time hackers“phishing attack” create loopholes in the system. In aphishing attack, step by step hacker gives allurement andtakes a few confidential information, which are essentialfor administrative control. Therefore, awareness andalertness of end‐user are extremely important.


SMART & ENERGY TECHNOLOGIES

FIGURE 1 The IoTEcosystem for Smart Building

Uncertainty in building performanceBuilding performance simulation is a widelyaccepted method to predict energyperformance during the design stage. Asimulation model helps to predict the energyperformance of a building and select the mostenergy efficient design strategies andalternatives. This is helpful to design new buildingsand/or retrofit old ones. However, largediscrepancies have been observed betweenpredicted and actual energy performance. This isattributed to several sources of uncertainty, e.g.,occupants’ activities, weather conditions, andmaterial properties. Figure 1(a) compares the EnergyUse Intensity (EUI, kWh/m^2) between simulated(predicted) energy consumption modelled during thedesign phase and the measured energy consumptionfor 62 LEED certified buildings in the United States.Figure 1(b) shows that in some cases, the actualmean value of material density would be far fromquoted value provided by manufacturers. Someresearchers found that thermal conductivity iscorrelated with material density and moistureconditions.

The geographical location and maritime exposuremake Singapore’s climate very humid (relativehumidity in the range 70‐80%). Thus, the moistureeffect on the uncertainty of thermal conductivity ofmaterials cannot be neglected.

In this research, a case study for a typical room inSingapore is considered. The floor, ceiling and threewalls are made of 150 mm concrete with two 20 mmcement sand plaster. In the remaining wall (eastfacing), two alternatives façades are considered: (i)similar to the other walls, traditional 150 mmconcrete façade, with two 20 mm cement sandplaster layers outside concrete layer, and (ii) phasechange material (PCM) façade, composed of one 10mm phase change material in the core, two 70 mmoutside concrete layers and two 20 mm cement sandplasters in the most outside layers. The describedPCM façade is expected to reduce the temperatureinside the room in the daytime and to release theheat in the night. An integrated approach based onmulti‐criteria decision making under uncertainty ispresented to select the optimal design alternative.

Multi‐criteria decision‐making platformTwo criteria have been considered: energyconsumption and economical cost (initial cost +equivalent annual cost). Energyplus V9.0 and Rpackage are used here to evaluate the value ofenergy consumption. Detailed insights into energyusage are considered. At first, sensitivity analysis hasbeen performed for all material parameters relatedto the façade. The most important variables havebeen screened according two different criteria: (i)their degree of uncertainty, and (ii) how they mayaffect the energy performance of the buildings(Mosalam and Alibrandi, 2018).

As a result, the following random variables havebeen selected with reference to the annual energyconsumption: thickness, conductivity, density andspecific heat of concrete layer in the concretefaçade, thickness of the PCM layer, thickness,conductivity, density and specific heat of the outsideconcrete layer and the inside concrete layer in thePCM façade.

For economical cost, the market unit price ofconcrete, PCM, and cement sand plaster areselected as random variables for the initial cost,which include installation fee. In order to considerboth the time value of money and energy saving costusing PCM façade compared with traditional façade,Equivalent Annual Cost (EAC) is adopted.

FAÇADE TECHNOLOGY


Yidan Gao, Umberto Alibrandi and Khalid Mosalam

FIGURE 1 (a) Measured versus design EUIs. (b) Spread in measured densities and their relation to quoted density for lightweight concrete blocks supplied by two different manufacturers

Multi-criteria Decision-making for Sustainable and Resilient Building Design under Uncertainties in Singapore

The case study results show that the criteria areconflicting. In terms of energy consumption, PCM isperforming better than concrete façade, asexpected (Figure 2). However, the initial cost ishigher especially considering a lifespan longer than35 years (Figure 3). Since after 35 years, there isanother initial cost for pcm façade, there is adiscontinuity in EAC of PCM in Figure 4. From theother side, after around three years, the EAC forPCM is lower than that for concrete, which meanslower lifecycle costs. For a lifespan of 50 years, thetwo alternatives have comparable cost (Figure 4),and thus PCM can be considered the best option.

ReferencesMosalam K.M., Alibrandi U., Lee H. & Armengou J. (2018).Performance‐based engineering and multi criteriadecision analysis for sustainable and resilient buildingdesign, Structural Safety, 74, 1‐13

Predicting thermal comfort is difficult. We recentlyshowed that the most used thermal comfort model,the PMV, predicted the thermal sensation correctlyonly one‐third of the time. One way to have betterpredictions could be to directly measurephysiological parameters using wearable sensors. Totest this hypothesis, 14 participants wore lab‐gradewearable sensors (e.g., heart rate, skin temperature,accelerometry) for two weeks (Figure 1) and self‐reported their thermal preference (cooler, nochange, warmer) as many times as possible everyday. Each day they wore the sensors for at least 20hours in order to capture their daily activitydynamics, which may not be covered in a controlledlab environment.

Using Wearable Sensors to Predict Thermal Comfort


AI AND MACHINE LEARNING

Shichao Liu, Stefano Schiavon, Hari Prasanna Das, Ming Jin, and Costas J. Spanos

FAÇADE TECHNOLOGY

FIGURE 2 Criteria 1: yearly energy consumption per total building area.

FIGURE 3 Criteria 2: Initial cost of façade systems

FIGURE 4 Mean value of EAC for two façade alternatives versus year


We related the thermal preference to the measuredparameters (Figure 2). We then built personal comfortmodels using machine learning. We found that thepredictive powers of the personal comfort models wereapproximately 24%/78%/79% (Cohen'skappa/accuracy/AUC), much higher than theconventional PMV and adaptive models as expected.However, the predictive power is highly dependent onthermal sensation. For example, the Cohen's kappa isonly 3% (0% means random guess) when the thermalsensation is from ‐0.5 to 0.5 as the general designpractice aims (Figure 3).

In the study, we also calculated the predictive powers ofdifferent features combinations because we wanted toexplore the trade‐offs between predictive accuracy andcosts related to those inputs collected. We found that theskin temperature measured at the ankle provides abetter prediction of thermal sensation than themeasurement taken at the wrist.

For more details on the paper: Shichao Liu, StefanoSchiavon, Hari Prasanna Das, Ming Jin, and Costas J.Spanos. "Personal thermal comfort models with wearablesensors." Building and Environment (2019): 106281.10.1016/j.buildenv.2019.106281

Open‐source version: https://escholarship.org/uc/item/3fb0p5gk

AI AND MACHINE LEARNING

FIGURE 3 Predictive power with different thermal sensations and classification methods

FIGURE 2 The distributions of thermal preference votes

FIGURE 1 Physiological sensors and wearing locations


Daylighting is one of many crucial aspects toconsider in the design of a building. When utilised inconjunction with controls and sensors, daylight canhelp reduce the energy consumed from artificiallights and promote levels of visual comfort andsatisfaction. Based on research, the building façadeenvelope can have a strong influence on both theartificial and natural light inside the occupied space. Tostudy these unique influences, SinBerBEST have setup aspecialised test‐bed chamber equipped with a fullyglazed façade and a controllable indoor lighting system.To emulate the behaviour of daylight inside the test‐bed,an array wall of light emitting diodes (LEDs) which can bevaried in both brightness and colour temperature is used.This system is known as the Daylight Emulator.

The Daylight Emulator array consists of eight wall panelscovering a total area of 27 m2 (9 metres in length and 3metres in height). Each wall panel consists of a mixtureof ‘cool’ and ‘warm’ LEDs (Figure 1), which are mountedin an alternating arrangement. Between the array wall ofLEDs and glazed façade, a membrane with diffusiveproperties is used to uniformly distribute the light acrossthe area of the window. The Daylight Emulator is capableof producing a maximum illuminance of approximately100,000 lux and colour temperatures ranging from 2400to 10,000 K, respectively. At its maximum output, theDaylight Emulator has a total power rating of 32 kW.

The Daylight Emulator can be programmed to simulate arange of different time conditions (i.e. sunrise or sunset)and sky conditions (i.e. a clear sky with no clouds). Alllighting controls that are linked to the Daylight Emulatorcan be accessed via a web‐browser user interface. AtSinBerBEST, the Daylight Emulator has been setup fortwo different test‐beds. This allows a wide range ofdaylighting‐based experiments to be conducted bySinBerBEST researchers. Due to the wide range offlexibility over the test‐beds, studies with differentresearch objectives have been conducted to investigate,respectively, indoor lighting controls, translucentconcrete panels, glare, and the influence of windowview.

There were a few studies conductedwhere the Day Light Emulator wasused extensively:

Indoor Lighting Control Development.In practice, light is delivered intobuildings so that it achieves a minimum level of light andthat it is uniformly distributed on a desk level surface.Since this light can originate from either artificial ornatural sources of illumination, the active use of daylightcan help reduce the reliance placed on electrical powerlights and thus save energy. Both test‐beds are equippedwith light (lux) meters that are placed on a regular grid,which are used to accurately measure the amount anddistribution of light falling on surfaces of interest. Theselux meters have been used in the development ofcontrols that can be used to reduce energy from indoorartificial lighting. When light from the Daylight Emulatoris received on a lux meter, a feedback signal will be sentto reduce the output of a nearby ceiling light in that areaof the room – reducing its output so that the target lightlevel is achieved. In other words, if the target level is 500lux and the Daylight Emulator provides 450 lux, then anearby ceiling light will be dimmed so that it provides theremaining 50 lux that is required. Conversely, should theincoming daylight excessively exceed the target lightlevel, a feedback signal will be sent to the roller blind(s)to reduce the harsh light received on the surface. Thesecontrols are also programmed so that when the light isbelow the target level, the roller blinds will be retracted.Because of the high degree of control over manyimportant parameters, the SinBerBEST team has beenable to develop and test the lighting controls under awide range of conditions in a relatively short period oftime.

Daylight Fenestration via Translucent Concrete PanelThe study of Translucent Concrete Panels (TCP) is one ofthe key areas of our research. This special concrete panelis embedded with multiple fibre optic tubes, which canbe used to replace conventional glazed materials. Thisallows the building façade envelope to maintain asuitable window‐to‐wall ratio, whereby the fibre optictubes act as the window to allow daylight to enter into

CYBER PHYSICAL SYSTEM

The SinBerBEST Daylight EmulatorI Komang Narendra

FIGURE 1 Left to right: Cool Light for clear blue sky, Mixture of cool and warm light for cloudy sky, Warm light for sun set and sun rise

12 – SinBerBEST sinberbest.berkeley.edu

the building. TCP have been installed in the façade of wallof a test bed and lux meters have been used to evaluatethe daylight performance. By varying the light intensityof the Daylight Emulator, tests are used to determinehow well light is transmitted through the TCP and intothe room.

Glare and Visual Comfort StudyThe Daylight Emulator can also be used to produceexcessive levels of light that causes a sensation of visualdiscomfort commonly known as glare. This is aphenomenon that often leads building occupants to useblinds to protect themselves against high levels ofbrightness from the window, and resultantly reduces theoverall levels of daylight inside the building. Anexperiment was designed using human participants,whereby they were asked to make self‐evaluations of

glare that they had experienced from the DaylightEmulator. The objective of the experiment was to testthe reliability of the predictive glare index formula usedto minimise glare from windows.

Besides allowing daylight to enter into the space,windows in buildings also contain dynamic visualinformation from the outside environment. This isdelivered to the occupants in the form of a view and is anessential part of any window design. The DaylightEmulator setup was modified by mounting photographicimages behind the glazed façade and using the LED arrayto backlight them. This gave the impression of a daylitwindow view, which allowed different types ofexperimental studies to be performed. Using 30 differentimages captured around the National University ofSingapore (NUS) campus, another study involving humanparticipants was performed. The overall aim of this studywas to evaluate the importance of having a window viewin the building.

CYBER PHYSICAL SYSTEM

FIGURE 2 TCP facade wall, lux meters measuring light from the Daylight Emulator

FIGURE 3 Glare study showing participant providing their self‐evaluations

FIGURE 4 Window view study showing the experimental setup (left) and participant viewing the window view image (right).

TECHNOLOGY NEWS


The Scientific Advisory Committee (SAC)meeting was held on August 6th, 2019 in CREATETower in Singapore. This meeting is held inaccordance with SinBerBEST’s research contractwith the National Research Foundation (NRF).Attendees in the meeting include the SAC members,Drs. Lim Khiang Wee and Adeline Lim and Md RafiOthman from NRF and Dr Gao Chun Ping from theBuilding Construction Authority (BCA). At this reviewmeeting, the theme leaders from the SinBerBESTprogram presented on their theme’s progressachievements and future plans. The SAC providedvaluable inputs to the program and recommendedavenues to be taken. The program thanks the SACfor their review and look forward to further successin future outcomes.

SAC meeting at the CREATE Tower.

Visiting FacultiesPawel WargockiProf. Pawel Wargocki, currentlyan Associate Professor in theDepartment of Civil Engineering,Technical University of Denmark,is an indoor climate scientist withover 20 years of researchexperience. His works havefundamentally influenced ourunderstanding of ventilation and indoorenvironmental quality’s impact on occupantperformance. His recent works have focused onhuman emissions and performance of sustainablebuildings.

As part of his ongoing collaboration withSinBerBEST, Prof. Wargocki was with us during the

second week of September. Theme A membershave been working with him on the design of acomprehensive study of the impact of elevatedlevels of carbon dioxide (CO2) on occupantperformance and physiological and psychologicalmarkers of cognitive effort. The rising atmosphericlevels of CO2 make this study very relevant andtimely. During this visit, the efforts were focused onfinalising the experimental design and on going oversome preliminary results obtained on physiologicaleffects of working in different CO2 concentrations.He also shared with the team his recent work onindoor environment benchmarking for buildings thathave undergone deep energy renovations, ALDRENTAIL. Collective discussions between Theme Amembers and Prof. Wargocki deliberated future,holistic examination of occupant comfort insustainable buildings.

Elliott GallProf. Elliott Gall is currently anAssistant professor at PortlandState University and the directorof the University’s Green BuildingResearch Lab. His researchinterests focus on physics andchemistry of indoor and urban airquality, human exposure to airpollution, and heat and masstransfer in the built environment.

Theme A has recently established a collaborativeventure with Prof. Gall and as part of this, he waswith us from the 28th of August to the 17th ofSeptember. During his stay, the primary focus wason carrying out a pilot study looking into how theburden of cognitive tasks changes human emissions.The study was carried out in the SinBerBEST climatechamber, over one week, with 16 humanparticipants. Prof. Gall collected air samples forlater, detailed spectrometric analysis of volatileorganic compounds emitted by the occupants whilethey were relaxed and while they were working oncognitive tasks. During his stay, he also planned withthe Theme members studies of personal exposureto indoor and outdoor air pollutants, to be carriedout in the near future.

Events and OutreachDr. Li Jiayu will be attending the Healthy Buildings2019 Asia and Pacific Rim (HB2019 Asia) conferencein October 22‐25, 2019 at Changsha, China.

SinBerBEST Scientific Advisory Committee Meeting

Professors Chandra Sekhar, Adrian Chong andClayton Miller attended and presented papers the2019 ASHRAE Annual Conference in Kansas City,Missouri, USA.

Professor Alberto Sangiovanni‐Vincentelli and Dr.Baihong Jin attended and presented papers in 2019IEEE International Conference on Prognostics andHealth Management in San Francisco and 2019 IEEEInternational Conference on Machine Learning andApplications in Boca Raton. Papers presented werein the topics of fault diagnosis and prognosis andmachine learning.

Professors Tham Kwok Wai, Stefano Schiavon andDr. Zuraimi Sultan provided invited speeches at the14th International Congress of PhysiologicalAnthropology in 24‐27 September at Singapore.

New StaffsArulmani Natarajan, Senior Instrumentation &Control Engineer

Ivanna Hendri, Design Engineer

Kavitha D/O Krishnasamy, Administration & HumanResource Executive

Koh Tsyr Harn, Senior Research Engineer

Sivasithamparam Karvannan, System Administrator

Interview with Prof. Elliott GallFor this issue, we talked with Dr Elliott Gall whois an assistant professor at Portland StateUniversity (PSU) in the department ofMechanical and Materials Engineering. ProfessorGall obtained his Ph.D. at the University of Texas,Austin. He was recently acknowledged with the 2018Yaglou Award from the International Society forIndoor Air Quality and Climate for his work on indoorozone chemistry. His research and teaching seeks toimprove the sustainability of the built environmentthrough an understanding of the intersection ofindoor air quality, urban air pollution, and humanexposure to air pollutants. We are fortunate to beworking with Dr Gall in Singapore again as acollaborator. In this interview, we took theopportunity to learn from Dr. Gall’s experiences withSinBerBEST.

Tell us about your area of research and why it is soimportant?I study indoor and urban air quality. My group doesexperiments and modelling to understand sources,sinks, and transformations of indoor and urban airpollution. We spend around 90% of our time indoors,and over half the world’s population lives in cities.There exist many air pollution sources andtransformation processes in both the urban andindoor environment that may adversely impacthuman health. This makes indoor air in buildings incities an especially important microenvironment forunderstanding air pollutant dynamics and airpollution exposure.

What made you joined SinBerBEST back in 2003?I joined SinBerBEST in 2013 for a few reasons. Thefirst was that it presented an opportunity to work ona new set of problems that were related to the areaof my Ph.D.: how to improve indoor air qualitywithout incurring a major energy penalty.SinBerBEST framed this question specifically for thechallenging case of the tropics, where buildingenergy use can be higher due to a year‐roundclimate with high heat and humidity. There was alsoa chance to work within a team of postdocs andfaculty, which included a number of well‐knownexperts in the field. Finally, my wife and I had alwayshoped to have an opportunity to live and workabroad, and to really experience a new culture.

What were your goals in SinBerBEST and did youachieve all of them?I had a various goals that were in some way relatedto my time as a postdoc in SinBerBEST: professionalgoals that were practical and specific, otherprofessional goals that were more abstract, as wellas personal goals. My practical goals were to ensure I

14 – SinBerBEST sinberbest.berkeley.edu

TECHNOLOGY NEWS

Prof Gall and Drs Li and Mishra preparing IAQexperiments in the test bed chamber

achieved outcomes that would build my CV andmake for compelling applications to facultypositions. That included publishing papers, improveproposal and academic writing, and initiating newresearch within SBB. I’d recommend that anypostdoc consider such practical outcomes, as mostpostdocs are, by definition, not intended to be long‐term employment. More abstractly, I had hoped togrow expertise in a few new analytical skills byleveraging resources and new collaborations inSingapore. Finally, my wife and I had personal goalsthat included traveling to other SE Asian countries toexperience life in the region, but while also settlingin to Singapore in a way that it felt like a secondhome.

After 3 years in SinBerBEST, what can you tell aboutyour impressions working in SinBerBEST and living inSingapore?I worked for almost 3 years in Singapore as apostdoc, and returned for 3 weeks in the summer of2019 to conduct experiments with SBB2. As apostdoc, it was a major life change, and took timefor me to adapt and adjust to a new place and whatwas, at the time, a new research institute. Inhindsight, I feel that Singapore can be an excellentplace to conduct research, but can also be a difficultplace to initiate research. The latter is, to an extent,true of anywhere. In Singapore, I get the sense thatthere are major funding initiatives that can provideintellectual freedom and opportunities that areunique. For example, it is exceedingly rare in the USto see funding opportunities like those provided tothe CREATE research centers and to centers at NTUand NUS. I expect that these investments will pay offfor Singapore for many years to come, as facilitieslike SBB’s testbed become published in theliterature and well‐known to researchers around theworld.

What do you hope to do next?I hope to continue to grow my own research groupat Portland State, where my lab, the Green BuildingResearch Lab (ww.pdx.edu/green‐building) conductsresearch on indoor and urban air quality. My hope isto focus on a few key research areas, includingconducting work focusing on better linking airpollution measurements, mitigation measures, andhealth outcomes. My group has spent the past fewyears doing field work, and I so am also ready toslow down on field work and spend a bit more timeon controlled lab experiments with a morefundamental focus.

What are you doing now with the currentSinBerBEST researchers?This past summer I joined Theme A researchers(Prof Schiavon, and Drs. Mishra and Li) to conductcontrolled experiments in the test bed concerninghuman emissions of VOCs under varying emotionalstates. The experiments were a success, and it wasvery exciting to see the test bed and have a chanceto take full advantage of the suite of capabilities thatI remember being worked on when I was a postdoc.Thus far, our data looks very interesting – we hopeto complete analysis and see the full story by theend of this calendar year. A big thanks to all the labstaff and researchers who helped make it possible.

We wish Professor Gall all the best.

TECHNOLOGY NEWS


SinBerBEST aims to deliver energy efficient

building technologies for the tropical built

environment, while optimising human comfort,

safety, security, and productivity within the

building. This interdisciplinary research project

is organised into five themes: A ‐ Human‐

Building Nexus; B ‐ Smart Technologies and

Resilient Buildings; C ‐Agile Design for Energy

Efficiency and Human Comfort; D ‐ Data

Analytics; and E ‐ Test Beds and Deployments

If you are interested to learn more about our

program, participate in our research or use our

test‐bed facilities, please contact Dr. Zuraimi

Sultan at zuraimi.sultan@bears‐berkeley.sg or

visit us at www.sinberbest.berkeley.edu

EDITORSCostas SpanosZuraimi Sultan

DESIGNSamuel Foo

CIRCULATIONKavitha D/O Krishnasamy

Please contact at kavitha@bears‐berkeley.sg

for any inquiry or further information.

CONTRIBUTORSUmberto Alibrandi Sergio AltomonteToby Cheung Hari Prasanna DasMegan DaweYidan Gao Ming JinMichael Kent Nishant Kumar Aleksandra Lipczyńska Shichao LiuKhalid MosalamI Komang NarendraSanjib Kumar PandaJovan PantelicKrishnanand Radhakrishnan Stefano SchiavonCostas Spanos

Fall 2019 Volume I

The SinBerBEST program, funded by the National Research Foundation (NRF) of Singapore, is aresearch program within the Berkeley Education Alliance for Research in Singapore (BEARS). Itcomprises of researchers from University of California, Berkeley (UCB), Nanyang TechnologicalUniversity (NTU) and National University of Singapore (NUS). SinBerBEST's mission is to advancetechnologies for designing, modeling and operating buildings for maximum efficiency and sustainabilityin tropical climates. This newsletter, published quarterly, is to showcase the excellence of SinBerBESTfaculty, post doctoral fellows and students.

Published by: SinBerBEST, Berkeley Education Alliance for Research in Singapore (BEARS) Limited1 Create Way, #11-02, CREATE Tower, Singapore 138602

Documents

IoT SENSING FOR BUILDING RESILIENCE ... › sites › default › files › sbb...Nishant Kumar, Krishnanand Kaippilly Radhakrishnan, Sanjib Kumar Panda The significance of the economic