Research Plan
Title
●
A concise summary of the proposed work, include appropriate detail
Abstract
●
●
●
A quick, one paragraph summary of the basic premise of your proposal.
Addresses the questions you will seek to answer, your need statement, and a brief description of relevance.
Hint: This is the last section you should write.
Background / Literature Review
Intro: Introduce your topic by identifying the big picture social issue you are addressing and explaining what is
currently known about the problem.
● Include a question, statistic, or phrase that will draw the audience’s attention. You need a hook that
draws the reader’s/audience’s interest. The hook directly relates to your project, or the open space in
your big picture.
● Explain why your topic should be of interest to the reader/audience. It would be best if you explain
how your topic has local/national/global implications.
● Include 1-3 figures/schemes if they help convey meaning and support your text (referenced properly,
of course).
● Hint: verbally explain the Context for your Project to 2-3 other people. This will help you organize the
information into logical bites.
Background: Provide the technical background necessary to understand the area you are working in.
● Identify and explain the technical concepts that are involved in your project. This is the most important
aspect of your research project. Think about teaching your audience by:
○ Explaining the concepts that show why your project is worth pursuing
○ Defining key terms related to your experiment
○ If helpful, include 2-3 figures or schemes. Be sure to reference the figures properly.
● Be sure that this section flows and does not feel like a laundry list of facts.
● Explain how these concepts relate to your project.
Lit Review: Provide a review of the work that has already been done in the field as well as clearly identifying the
specific open space for your work.
● For a review of existing work, identify 2-3 published studies closely related to your project area. For
each of the related studies, complete these steps in the following order.
○
○
○
○
Briefly describe the experiments findings
Describe the experiments’ observables.
Give results from the experiment and what those results mean (as they relate to your
proposed project area)
Include a description of the “open space” that sets up your project.
Problem Statement
●
●
What specific problem are you addressing or working to solve? This will be your need statement. By this time in
the writing, you have told the story
Context → Background → Current Work → Open Space → Need Statement
and the reader should truly appreciate the scope of the problem and why you want to solve it.
What is your proposed solution (device, process, etc.)? This is the overview of your project. The details are in the
next section.
Project Planning
●
Identify and describe the project details that support the completion of your solution.
●
●
●
○ Describe your successful proof of concept
○ What additional hardware, software, integration, etc. will lead to a viable solution.
List the resources that you will need to complete the project.
○ This should include hardware, professional support, time, money, etc.
○ This section is the basis for a funding proposal.
Provide a work plan (include a task list plus associated timeline).
○ The work plan should be sufficiently comprehensive to engender confidence that significant planning
details have not been overlooked.
○ Include a project specific Gantt chart, making sure that it aligns with the course Gantt & timing.
○ Clearly identify the critical path (project component sequencing and which steps are dependent on
others)
Describe your validation plan.
○ What measurable data will you use to evaluate the efficacy of your project?
○ How (with what timing and what tools) will you acquire this data?
○ What analysis will be required in order to claim a successful project?
Works Cited
●
●
All citations must be APA format.
Be sure that when referencing these works above, proper in-line citations are used.
Chalk Talk - on demand
A 5-10 minute presentation of your research at this point. The first couple of items should be well defined, and you should
be comfortable discussing these conversationally. Items 3-5 will change as the project develops.
1. Project description - 1 sentence
2. What is the social connection
3. What have you done
4. What are you doing now
5. What do you still need to do
Introduction
The world is quickly urbanizing, and a key problem is that there is inadequate living and
working spaces. To deal with the increasing number of automobiles, most cities in China and
other crowded cities worldwide tend to save more space by turning outdoor parking or parking
structures under the ground. This not only saves places for the government but also saves money
for the constructors that they don’t have to buy additional lands for parking anymore.
Because of the rapid growth of underground construction, it is essential for people to
realize the danger they put themselves into every day, and they can’t really avoid it. Even though
some people might live in a house that they can have their car park outside on the street instead
of a garage, in commercial centers and their workplaces, owners tend to like their parking
underground to use the land most effectively.
How often do you go to the underground parking lot or stay in any indoor spaces, like an
office without a window or subway stations? And how long you usually stay there? Most people
remain inside the building longer than they think, and most of them don’t know about the
damage from indoor pollution. “The major pollutants emitted by motor vehicles including CO,
NOx, sulfur oxides, (SO), HC, lead (Pb) and suspended particulate matter (SPM), have damaging
effects on human health”(Bhandarkar). They can cause harmful diseases like asthma, heart and
lung issues, increase the risk of getting cancer, and premature death. All this brings up the
importance of ventilation systems in indoor spaces. By modeling the Ventilation system, people
can know how sufficient ventilation is and whether there is a better or cheaper way to perform
the same performance. (reword)
-
Next paragraph
-
discuss parking structure design and ventilation research
Problem statement
The people who are indoors a lot and park underground need to know the effectiveness of
the ventilation system by testing different models exist on the market, so as to avoid the damage
brought by indoor pollutants.
Reference
Bhandarkar, S. (n.d.). Vehicular Pollution, Their Effect on Human Heatlh and Mitigation
Measures. Retrieved from
https://d1wqtxts1xzle7.cloudfront.net/48231041/VE004_1_2_33_40.pdf?1471849302=&respons
e-contentdisposition=inline%3B+filename%3DVehicular_Pollution_Their_Effect_on_Huma.pdf&Expires
=1612498834&Signature=RymQIMiAi17m1KcUT9LnhOcIk3p8uvsdH7n9MXD0heYkCXOgI1
TBCgiBOoRx0OGXedaBLVTFk2qAVfyrD7CWUHB~aNetLmoJyTI9cHSDktYO24iUgr7Ckjts
X~C974LgUT8sMEsvqHiFxdag6Ra8n7gCMXkubciQ8rFRgt6cU~GIiTQ6RvjzTJhjFpP5J5IPwlhS-rpeZhV6GaKKD71nmm3zvVoabxJwq3BBUhrhIGFUefC0qiwpysh7caOgcFsMTtfUxM7OjDqGL267oEMqV2
ON0iKOGOnNv6fDa7dAqwe9GEqhGybEw1n1u9PV2cRn90J8cY1TNf1xr-MxGRoA__&KeyPair-Id=APKAJLOHF5GGSLRBV4ZA.
https://www.pnas.org/content/111/23/8386.abstract?etoc
https://www.sciencedirect.com/science/article/abs/pii/S036013231730358X
Read article on HVAC in garages
http://blog.dwyer-inst.com/2020/03/04/understanding-ventilation-efficiency-and-dcv-in-parkingstructures-and-garages/#sthash.piGTBDu4.dpbs
What is building code for China?
● https://www.gbpn.org/sites/default/files/China_Country%20Summary_0.pdf
https://scihub.wikicn.top/10.1016/j.apr.2020.08.008
Indoor air quality investigation at air-conditioned and non-air-conditioned markets in Hong
Kong https://doi.org/10.1016/j.scitotenv.2003.09.031
2020 Kim Paper --Exposure to respirable particles and TVOC in underground parking garages
under different types of ventilation and their associated health effects; Oh, HJ., Sohn, JR., Roh,
JS. et al. Exposure to respirable particles and TVOC in underground parking garages under
different types of ventilation and their associated health effects. Air Qual Atmos Health 13, 297–
308 (2020). https://doi.org/10.1007/s11869-020-00791-0
I have only intro here. I need a background section. The background is used to educate the reader so they
can understand my project. I will need sections as follows:
1. Air pollutants and health
2. Measurement devices
3. Parking structure types
4. ventilation systems
(if you don’t have access to any article, use sci-hub)
You don’t have to use all the articles. If you find better replacement, put them on. Follow the example and
the guideline I send you.
Asmar 1
Kristen Asmar
Honors Engineering for Social Good
Research Proposal
Identifying Fatigue in the Vastus Lateralis Muscle in Athletes Recovering from Knee
Injury
Abstract
Athletes overuse their muscles when doing rehabilitation exercises in an attempt to
shorten their rehabilitation process and return to their exercises or sports. In order to prevent
overusing muscles, this device warned users when his/her muscle is fatigued and therefore to
stop rehabilitation for that day. The hinged brace told the user when he/she has reached the
desired angle for hamstring curls and when he/she has held his/her knee for the desired amount
of time. For validation, the brace was tested on someone who tires easily, someone who was
recovering from a knee injury–making sure they were well into their rehab in order to prevent
additional damage when testing–or on a healthy person by inducing fatigue. For a control, the
knee brace was tested on someone who is not recovering from a knee injury. In the future, the
brace can be modified for active rehabilitation, controlling the patient’s knee for him/her and
stopping the movements when his/her muscle is fatigued.
Intro
In an effort to shorten their rehabilitation period and improve their injury faster, athletes
over do physical rehabilitation. There is a misconception that more rehabilitation will equal more
recovery, so atheltes perform their rehabilitation exercises longer than necessary (Hilliard,
Asmar 2
Robert C, et al, 2017). As a result, the athletes overuse their recovering muscles and cause more
damage to their injuries by tearing the damaged ligaments more or by halting the progress due to
damaging the muscle after having rehabilitated it some.
This device will use electromyography, encoders, and LEDs as a warning system for
muscle fatigue. The millivolt output of the EMGs, which is based on amplitude, will be inputted
into algorithms to calculate the root mean square and median in order to take the raw EMGs and
transform them into quantitative data. The EMGs will be taken through skin sensors placed on
the vastus lateralis muscle, which is on the outer side of the thigh. Once the root mean square
increases and the median frequency decreases to a certain amount, the user will be warned that
his/her muscle is fatigued. If time allows, encoders will be added to the brace to let the users
know when they have reached the desired angle for hamstring curls. The additions to test all of
this will be added to a neoprene knee brace with hinges, making sure to keep the knee brace light
and mobile.
Background:
After talking with the assistant athletic trainer at the University of La Verne, Meg Ryan,
it was decided there was a gap in rehabilitation for athletes. It is important to receive feedback
about the muscles that athletes are trying to recover because it can lead to more damage to the
already hurt muscles. It would be beneficial to receive biofeedback for how much the muscle is
firing so as to get a better grasp at how well the muscle is recovering from surgery or injury
(Ryan). In a gait training study done on children with cerebral palsy, the group who received
biofeedback during the training did statistically better in regards to muscle movement (Dursun,
Dursun, Alican, 2003), suggesting that biofeedback is favorable and important in therapy.
Asmar 3
Therefore, it would be beneficial to integrate sEMG sensors into the knee brace so that raw EMG
signals can be converted to root mean squares (RMS) and median frequencies (MF) in order to
muscle firing information.
Electromyography, or EMGs, is the technique used to record and analyze the electrical
activity of skeletal muscles. EMGs are useful when determining how healthy muscles and motor
neurons are. The EMG itself is a measurement of the electrical speed and strength between two
points on the body, in response to nerve stimulation, such as contracting the muscle. sEMGs are
surface EMG measurements that use electrodes that stick directly onto the skin in order to
measure muscle electrical activity, as opposed to needle EMGs, which are inserted directly into
the body. EMGs can be used to determine when muscles are fatigued through MF and RMS. A
fatigued muscle caused MF, the frequency value that consists of two parts of equal energy that
separates the power spectrum, which can be determined through algorithms within code. RMS is
value after averaging the amplitudes of the raw EMG signals. When the median frequency
lowers and the value of the root mean square increases, the muscle is fatigued.
Lit Review #1: Development of an EMG-Controlled Knee Exoskeleton to Assist Home
Rehabilitation in a Game Context
Mingxing Lyu, Wei-Hair Chen, Xilun Ding, Jianhua Wang, Zhongcai Pei, and Baochang Zhang
made an EMG exoskeleton knee rehabilitation device for patients who are in the initial phases of
recovery after a stroke. Previous rehab exoskeletons required assistance from a physical therapy
and can cost patients a large amount of money to use, so this device was made to be usable at
home so as to lower the price of rehab. Also, this device motivates the user to complete rehab
exercises because of the game attached. In a study done on stroke patients performing hand
rehabilitation through a gamified approach, therapists and patients provided positive feedback to
Asmar 4
gamified rehabilitation processes because of an increase of interest and motivation to complete
the rehab exercises (Carneiro, et al., 2018). The creators modified the app Flappy Bird so that the
patients’ knee movements control the height of the bird. To translate a knee movement to flight
of the bird, the device uses a Myo thigh-band, which consists of 16 dry surface EMG sensors and
is easily adjustable. The Myo thigh band has a sampling frequency of 200 Hz and sends raw
sEMG data to the host computer via Bluetooth Low Energy. A Kalman filter is used to calm
down the noisiness of raw EMG data. The exoskeleton itself assists the patients with both hip
and knee movements and is made out of aluminum and 3D-printed parts so as to limit the weight
of the exoskeleton.
Lit Review #2: A Comparative Study of EMG Indices in Muscle Fatigue Evaluation Based on
Grey Relational Analysis during All-Out Cycling Exercise
Lejun Wang, Yuting Wang, Aidi Ma, Guoqiang Ma, Yo Ye, Ruijie Li, and Tianfeng Lu studied
different EMG indices in muscle fatigue in all-out cycling exercises in order to determine which
EMG calculation was most reliable for detecting muscle fatigue in cyclists. Muscle fatigue is
identified through a reduced capacity to generate force or create power output. Grey relational
analysis was used to quantify the change of trends in identified sequences by using grey
relational grade. The EMG sensors were placed on the vastus lateralis muscle, as it the muscles
most responsible for power in cycling exercises. The indices studied were EMG root mean
square (RMS, median (MF), and mean power frequency (MPF) based on Fourier Transform. 10
volunteer cyclists completed a 5 minute warm up, 3 minute rest, then a 30 second high power
session. The scientists found that the MF and MPF were the more reliable than EMG RMS when
determining when the muscles become fatigued, which was when the MF and MPF showed a
significant decreasing tendency. But, the MF and MPF are limited by the Fourier Transform,
Asmar 5
wavelet transformation has been observed as an alternate when dealing with more dynamic
contractions. Overall, MF derived from wavelet transformation was the most reliable EMG
indice when determining muscle fatigue because it had the highest grey relational grade.
Lit Review #3: A Muscle Fatigue Monitor Based on the Surface Electromyography Signals and
Frequency Analysis
Patients completing rehabilitation may experience muscle fatigue or may further damage their
injury without knowing. In order to alert the user of these unknown pains, Gutiérrez, Cardiel, and
Hernández created a way to monitor when muscle fatigue is present. The two factors to
determine when the muscle was fatigued are: a shift of the median frequency in the power
spectrum density and the Borg scale criteria related to pain. EMG signal processing was used to
determine when the median frequency shifted and the root mean square value was measured
within the time domain in order to determine when the muscle is fatigued. Both the values for
median frequency and root mean square were determined by using algorithms programmed in
LabVIEW. The muscle was determined to be fatigued when the value of the median frequency
decreases, while the root mean square values increase, meaning the frequency of signals of
sEMGS, but the amplitudes of the signals increased, reflecting the overexertion of the muscle.
The rehab exercise I will focus on is hamstring curls. The exercise consists of the patient
holding on the back of a chair and bending their knees towards their back as far as they could.
The patient should hold their leg for 5 seconds once they reach the angle. The process I will
create will make sure the patient completes 3 sets of 10 hamstring curls. After each set, the knee
brace will encourage the patient to bend their knee more. Hamstring curls were chosen because it
is a popular rehabilitation exercise for recovering athletes, according to the American Academy
Asmar 6
of Orthopaedic Surgeons, and it will mostly be used as a proof of concept exercise because it
involves angles and movement of the knee.
Driving Questions, Hypothesis, Open Space for New Work
Driving Question: Will this device prevent the overuse of muscles?
Hypothesis: If muscle fatigue is detected in the vastus lateralis muscle of athletes recovering
from knee injury, then the athlete may recovery more fully due to limiting further injury during
the rehab process.
Open Space for New Work: In the future, the knee brace can be improved by adding a
mechanical controlling device that raises the knee for the patient when the patient is not ready for
active rehabilitation. The knee brace would still detect when the muscle is fatigued and when to
stop the knee from bending, but the knee brace would be the device controlling and lifting the
knee, rather than the patient.
Methods/Materials
Methods:
The proposed device is a knee brace that detects fatigue in the vastus lateralis and the
desired angle for hamstring curls. The myoware sensor and EMG skin sensors will be used to
measure the EMG signals of the muscle. For fatigue detection, RMS and MF will be calculated
using the algorithms found in the study, A Muscle Fatigue Monitor Based on the Surface
Electromyography Signals and Frequency Analysis (Gutiérrez, et al., 2016). The raw data from
Asmar 7
the EMGs is the millivolt output of the signals based on amplitude. Those millivolt values will
be inputted into the algorithms, which will generate a numerical value that will be analyzed,
possibly every second. When the RMS and MF reach a certain point, the RMS increasing and the
MF decreasing enough, the brace will notify the user that the muscle is fatigued with an audible
noise from an Arduino speaker.
If time allows, encoders will be used to detect and control the angle at which the knee is
bent. LEDs will be used to warn the user when he or she has reached the desired angle by using
green, yellow, and red lights, so that there will be no learning curve. Another LED will be used
as a timer, blinking every second, so the user will know how much time has passed since
reaching the desired angle.
The knee brace will be coded using Arduino and the additional parts will be added to an
already functioning knee brace. The knee brace chosen is a combination of a neoprene sleeve and
hinges, in order to stabilize the knee enough during the exercise and have hinges available if
encoders can be added. If live data acquisition is not possible due to lack of coding proficiency,
then a backup plan is to download the data to be later analyzed by a therapist. The injured person
will go through the entire rehabilitation protocol while using the knee brace. At the end of the
protocol, the data will be downloaded so that the therapist can note when or if fatigue was
reached. This knowledge can advise the therapist on if adjustments to the patient’s rehabilitation
protocol are necessary.
For validation, I will test the device on a person who tires easily, is recovering from a
knee injury, or on a healthy person. The recovering athlete and the healthy person are the most
plausible clients because they provide their own controls. For the athlete, the control is the noninjured knee, and for the healthy person, the control is data acquisition of the muscle that was not
Asmar 8
forced to fatigue. If the healthy person is the chosen client, then in order to test the fatigue
detection, the muscle of the client needs to be forced into fatigue through prolonged knee
movement or short bursts of exercise.
Materials:
Knee braces
1. https://www.shockdoctor.com/products/ultra-knee-support-with-bilateralhinges?variant=14354302435381
a. $69.99
Arduino Nano
1. https://store.arduino.cc/usa/arduino-nano
a. $22.00
Arduino speaker
1. https://www.adafruit.com/product/1890
a. $1.95
Arduino
1. https://store.arduino.cc/usa/arduino-uno-rev3
a. $22.00
MyoWare sensor (detects EMGs)
1. https://www.adafruit.com/product/2699
a. $37.95
3M Red dot EMG surface sensors (100)
Asmar 9
1. https://www.amazon.ca/Red-Dot-Multi-Purpose-MonitoringElectrode/dp/B01AME7YC0/ref=pd_sim_147_1/136-88739542404105?_encoding=UTF8&pd_rd_i=B01AME7YC0&pd_rd_r=c7d59e69-ec62-4e4d8ccd-7e2385b8ca76&pd_rd_w=JxHMQ&pd_rd_wg=f54pn&pf_rd_p=ca0a769f-467a4dd2-b2dc9f16d6aac0a1&pf_rd_r=QR7R55TVXNPB6TT1G0J7&psc=1&refRID=QR7R55TVXN
PB6TT1G0J7
2. $23.91
Timeline
Works Cited:
Asmar 10
American Academy of Orthopaedic Surgeons. Knee Conditioning Program. (n.d.). Retrieved
December 19, 2019, from https://orthoinfo.aaos.org/globalassets/pdfs/2017rehab_knee.pdf.
Carneiro, F., Tavares, R., Rodrigues, J., Abreu, P., & Restivo, M. T. (2018 ,November). A
Gamified Approach for Hand Rehabilitation Device. Retrieved from University of Port
website: https://onlinejour.journals.publicknowledgeproject.org/index.php/ijoe/article/viewFile/7793/4788
Cifrek, M., Medved, V., Tonkovíc, S., & Ostojíc, S. (2009, January).
Surface EMG based muscle fatigue evaluation in biomechanics. Retrieved from Clinical
Biomechanics website:
https://www.researchgate.net/profile/Mario_Cifrek2/publication/24200364_Surface_EM
G_Based_Muscle_Fatigue_Evaluation_in_Biomechanics/links/59f7759d0f7e9b553ebee4
6b/Surface-EMG-Based-Muscle-Fatigue-Evaluation-in-Biomechanics.pdf
Copaci, D., Martín, F., Moreno, L., & Blanco, D. (2019, March). SMA Based Elbow Exoskeleton
for Rehabilitation Therapy and Patient Evaluation. Retrieved from IEEE website:
https://ieeexplore.ieee.org/document/8658083
Dursun, E., Dursan, N., & Alican, D. (2003, September). Effects of biofeedback treatment on gait
in children with cerebral palsy. Retrieved from Taylor and Francis website:
https://www.researchgate.net/profile/Erbil_Dursun2/publication/8966208_Effects_of_bio
feedback_treatment_on_gait_in_children_with_cerebral_palsy/links/0deec524d23761ca5
a000000Effects-of-biofeedback-treatment-on-gait-in-children-with-cerebral-palsy.pdf
Guo, S., Ji, J., & Tao, S. (2014, December). Lower limb rehabilitation robot for gait training.
Retrieved from ResearchGate website:
Asmar 11
https://www.researchgate.net/profile/Shuai_Guo17/publication/278395290_Lower_limb_
rehabilitation_robot_for_gait_training/links/561335d708aea9fb51c28ff5.pdf
Hilliard, R. C., Blom, L., Hankemeier, D., & Bolin, J. (2017, May). Exploring the Relationship
Between Athletic Identity and Beliefs About Rehabilitation Overadherence in College
Athletes. Retrieved December 19, 2019, from National Center for Biotechnology
Information website: https://www.ncbi.nlm.nih.gov/pubmed/27632827
Krebs, H. I., Ferraro, M., Buerger, S. P., Newbery, M. J., Makiyama, A., Sandmann, M., . .
Hogan, N. (2004, October). Rehabilitation robotics: pilot trial of a spatial extension for
MIT-Manus. Retrieved from BioMed Central website:
https://jneuroengrehab.biomedcentral.com/track/pdf/10.1186/1743-0003-1-5
Lyu, M., Chen, W.-H., Ding, X., Wang, J., Pei, Z., & Zhang, B. (2019, August 27). Development
of an EMG-Controlled Knee Exoskeleton to Assist Home Rehabilitation in a Game
Context. Retrieved December 19, 2019, from Frontier in Neurorobotics website:
https://www.frontiersin.org/articles/10.3389/fnbot.2019.00067/full
Mayo Clinic. (n.d.). Electromyography (EMG). Retrieved December 19, 2019, from Mayo Clinic
website: https://www.mayoclinic.org/tests-procedures/emg/about/pac-20393913
Narang, Y. S., Arelekatti, V.N. M., & Winter, A. G. (2015, July). The Effects of Prosthesis
Inertial Properties on Prosthetic Knee Moment and Hip Energetics Required to Achieve
Able-Bodied Kinematics. Retrieved from IEEE website:
https://ieeexplore.ieee.org/document/7155597
Wang, L., Wang, Y., Ma, G., Ye, Y., & Lu, T. (2018, April 16). A Comparative Study of EMG
Indices in Muscle Fatigue Evaluation Based on Grey Relational Analysis during All-Out
Asmar 12
Cycling Exercise. Retrieved December 19, 2019, from Hindawi website:
https://new.hindawi.com/journals/bmri/2018/9341215/
Wu, Q., Wang, X., Chen, B., & Wu, H. (2011, October). Design and Fuzzy Sliding
Mode Admittance Control of a Soft Wearable Exoskeleton for Elbow Rehabilitation.
Retrieved from IEEE website: https://ieeexplore.ieee.org/document/8489867
Zhi, Y. X., Lukasik, M., Li, M. H., Dolatabadi, E., Wang, R. H., & Taati, B.
(2017, December). Automatic Detection of Compensation During Robotic Stroke
Rehabilitation Therapy. Retrieved from IEEE website:
https://ieeexplore.ieee.org/document/8214256
Environmental Pollution 247 (2019) 626e637
Contents lists available at ScienceDirect
Environmental Pollution
journal homepage: www.elsevier.com/locate/envpol
On-site assessments on variations of PM2.5, PM10, CO2 and TVOC
concentrations in naturally ventilated underground parking garages
with traffic volume*
Zhijian Liu a, *, Hang Yin b, Shengyuan Ma a, Gaungya Jin a, Jun Gao c, **, Wenjun Ding d
a
Department of Power Engineering, North China Electric Power University, Baoding, Hebei, 071003, PR China
Department of Civil Engineering, Technical University of Denmark, DK-2800, Kgs, Denmark
School of Mechanical Engineering, Tongji University, Shanghai, 200092, PR China
d
Laboratory of Environment and Health, College of Life Sciences, University of Chinese Academy of Sciences, No.19A Yuquan Road, Beijing, 100049, China
b
c
a r t i c l e i n f o
a b s t r a c t :
Article history:
Received 5 November 2018
Received in revised form
14 January 2019
Accepted 23 January 2019
Available online 25 January 2019
There have been an increasing number of automobile vehicles in cities, so that newly developed residential areas are mostly designed with underground parking garages (UPGs). For naturally ventilated
UPGs, the ventilation performance may be insufficient to discharge totally vehicle-induced pollutants out
of the enclosed underground spaces, which consequently results in threats to residents' health. This
study, therefore, aims at examining the patterns of pollutant concentrations in naturally ventilation UPGs
as well as their sensitivities to traffic volume. In particular, the naturally ventilated UPGs’ weekday
particulate matters (PM2.5 and PM10), CO2 and TVOC concentration as well as their relationships between
traffic volume were quantitively evaluated based on field measurements in eight residential areas in
Baoding, China. Results indicated that daily average PM2.5, PM10, CO2 and TVOC concentrations in studied
UPGs were 105.81 mg/m3, 464.17 mg/m3, 571 ppm and 24 ppb, respectively. The PM2.5 concentrations in
UPGs were slightly higher than that in ambient environments, while the PM10 concentrations in UPGs
were significantly higher. Furthermore, both PM10 and TVOC concentrations in UPGs were in significant
relationships with traffic volume at the p < 0.01 level, while the concentration of UPG PM2.5 generally
exhibited a significant correlation (p < 0.01) with that of the ambient. Nevertheless, a combination of
traffic volume, the ambient and accumulative effect was much better to explain the hourly PM10 concentration in UPGs. These findings will be conducive to instruct engineers with fundamental knowledge
of UPG ventilation design.
© 2019 Elsevier Ltd. All rights reserved.
Keywords:
Underground parking garage
Natural ventilation
Indoor pollutant concentrations
Daily variation characteristics
Correlation analysis
Multiple regression model
1. Introduction
Significant growth of automobile number has been witnessed
for the last few decades in China. Therefore, the construction of
underground parking garage (UPG) is popular for newly-built living
quarters due to the demand for parking space. Due to the severe
energy situation and the worldwide wave of urban sustainability
(He et al., 2018), most underground garages still use natural
ventilation systems. Thus, the air quality of UPGs under natural
*
This paper has been recommended for acceptance by Dr. Haidong Kan.
* Corresponding author.
** Corresponding author.
E-mail addresses: zhijianliu@ncepu.edu.cn (Z. Liu), gaojun-hvac@tongji.edu.cn
(J. Gao).
https://doi.org/10.1016/j.envpol.2019.01.095
0269-7491/© 2019 Elsevier Ltd. All rights reserved.
ventilation has become a hot spot. Compared with the temperature,
humidity, particulate matter (PM) and biological particles that are
often studied in previous research of indoor pollutants (Liu et al.,
2018a; Liu et al., 2018b), it is necessary to consider the impact of
traffic volume on the indoor pollution of UPGs. The pollutants
including carbon monoxide (CO), carbon dioxide (CO2), nitrogen
dioxide (NO2), particulate matter (PM), volatile organic compounds
(VOCs), etc. generated from automobiles would be concentrated
and accumulated in this enclosed or semi-enclosed micro-environments. Although there is no specific guideline or standard on
the air quality of UPGs, it is possible to quantify the health risks of a
large part of pollutants presented in garages, which including 39
compounds (Glorennec et al., 2008). Those who work in UPGs, such
as security staff and cleaners, would stay for a long time in this
environment. While car park users only enter the garage 2 to 3
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
times a day and the duration time of staying in garage is very short,
but each time they exposure to high concentrations of air pollutants. Therefore, exposure to traffic-related air pollutants in UPGs
would cause adverse health effects to both car park workers and
users (Bowatte et al., 2018; Costa et al., 2017; Li et al., 2014; Tham
et al., 2018; Zhang et al., 2017).
Many researchers have performed studies on the air quality of
UPGs, and research methods include numerical simulation and
field measurement (Chan and Chow, 2004; Jokl, 2000). For
example, Papakonstantinou et al. developed a CFD model for the
simulation of CO levels inside a typical central garage under natural
and mechanical ventilation, respectively, to investigate the ventilation effectiveness(Papakonstantinou et al., 2003). Song et al.
simulated particle concentration distribution under transient natural wind through the entrance of the UPG used three-dimensional
numerical model(Song and Zhao, 2017). Zhao et al. investigated the
transient air flow, particle dispersion, and particle exposure risk
during and after transient vehicle exhaust in an UPG(Zhao and
Zhao, 2016). In addition, most of studies on air quality of UPGs
were conducted on field measurement. Vukovi et al. compared the
instrumental measurements with active moss biomonitoring in
terms of the measuring accuracy of heavy metals and PAHs concentrations in PM10 samples collected in four parking garages. The
parking facilities selected as the research object were different in
size, structure, the number of parking spaces and the type of
ventilation system(Vukovic et al., 2014). Demir et al. measured the
air pollutant concentrations in terms of exhaust emissions of vehicles in multistory car parks under different ventilation system
including natural, ductwork and jet fan ventilation systems, and
determined that short-term exposure of humans to the environment of this car park would not cause health problems(Demir,
2015). Mar
c et al. measured the BTEX compounds concentrations
of a two-level UPG(Marc et al., 2016), while Hun et al. investigated
the indoor BTEX and MTBE concentrations in residences with an
attached garage, a detached garage or carport for natural ventilation(Hun et al., 2011). The results indicated that gasoline-related
VOCs from parked cars especially in attached garages would deteriorate indoor air quality. Batterman et al. also measured the VOC
concentrations at 15 residential garages during 4-day measuring
period and concluded that VOC concentrations were not strongly
correlated to air exchange rates (AERs), but appeared largely
dependent on door opening and VOC sources(Batterman et al.,
2006). Zhao et al. investigated the seasonal patterns of PM10,
PM2.5 and PM1.0 in a naturally ventilated UPG, concluding that the
natural ventilation could not guarantee the quality of indoor air in
this environment because the daily average PM10 and PM2.5 in the
garage generally exceeded the long-term (24 h) exposure limit of
the Chinese standard especially in winter(Zhao et al., 2017).
Moreover, the Pearson correlation and a mass-balance model were
employed to determine the impacts of traffic flow, AER and outdoor
particles. The relationship between indoor pollutants and traffic
volume has been found in many studies(Ho et al., 2004; Hun et al.,
2011; Zhao et al., 2017). Moreover, accumulative effect of indoor
pollutants was taken into consideration in the study of Ho et al.
(2004). Kim et al. evaluated the effect of traffic volume on air
pollution levels used a multiple regression longitudinal model
containing temperature, relative humidity, wind speed and traffic
activity (Kim et al., 2007).
Because of the high density of urban population, there have
been an increasing number of high-rise buildings in China. Meanwhile, the quantities of own cars are increasing by a big margin in
residential quarters, so that newly developed residential areas are
mostly designed with UPGs to solve the problem of parking space
shortage in residential areas. Therefore, the phenomenon of
widespread UPGs in densely populated areas has become common
627
in China. Moreover, for most residential UPGs in China, natural
ventilation is the most popular ventilating mode, considering
construction costs and operating costs. Only a few residential UPGs
are designed with mechanical ventilation, but the ventilation system dose not operate throughout the year. Additionally, there is no
study aimed at Baoding region (Hebei, China), which the value of
Air Quality Index (AQI) often ranked the top 5% among 338 cities in
China. Therefore, it is the first time for us to investigate the air
quality of residential UPGs under natural ventilation in Baoding
city, China. The air pollutant concentrations in parking facilities are
affected by many factors such as traffic volume, ambient condition
and the construction design of UPGs. However, only a few studies
have focused on systemically quantifying the influence factors of
indoor air quality.
In our study, field measurements were conducted to gain
various data of air pollutant concentrations and traffic volume in
eight naturally ventilated UPGs in the urban area of Baoding, China,
aiming at examining the patterns of pollutant concentrations in
naturally ventilated UPGs as well as their sensitivities to traffic
volume. In particular, the naturally ventilated UPGs’ weekday particulate matters (PM2.5 and PM10), CO2 and TVOC concentration as
well as their relationships between traffic volumes were quantitively evaluated by SPSS 16.0. Moreover, the ambient condition and
accumulative effect were introduced into multiple regression
model to explain the hourly PM10 concentration in UPGs. Furthermore, the over standard rates of indoor pollutant concentrations
based on both international and Chinese national standards were
also presented in this study.
2. Methods
2.1. Sample description
Eight single-layer UPGs with natural ventilation were selected
as measurement objects in this study, and all the samples are
located in the central of Baoding city (38.85 N, 115.48 E). The size,
structure, the design of air vent, the number of parking spaces and
the traffic volume were different in each UPGs. General information
about all the measured UPGs was listed in Table 1. It is widely
known that the air quality of the enclosed environment heavily
depends on the effectiveness of its ventilation system. While for
naturally ventilated UPGs, the air quality is mainly affected by its
characteristics containing the volume, number, size, as well as the
position of air vents and entrances/exits. These detailed information was also listed in Table 1.
2.2. Measurements
The field measurements were conducted on weekdays from
March 1st to April 10th, 2018, and the measuring time of each
sampling site was from 0:00 a.m. to 24:00 p.m. All measurements
were taken during weekdays because there are apparent morning
and evening rush hours in nearly all the residential UPGs. The data
of outdoor particle matter concentration (PM2.5 and PM10) were
directly collected from local meteorological environment monitoring stations, with the time interval of 1 h. The locations of
sampling sites and all the meteorological environment monitoring
stations in Baoding city were shown in Fig. 1, and the distance
between the sampling site and the nearer monitoring station was
listed in Table 1. For indoor parameters, the temperature, relative
humidity, PM2.5, PM10, CO2 and TVOC concentrations were
measured and the measurement range and accuracy was listed in
Table 2. The instruments were installed at position 1.5 m above the
floor in the main parking space away from the entrance or exit. All
data were collected continuously (during 24 h) in each sampling
628
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
Table 1
General information about the all the measured underground parking garages.
Case No.
Area (m2)
Height (m)
Volume (m3)
Parking space
Entrance/Exit
a
N
b
Air vent
Monitoring station
S1 (m)
d
P
N
c
Distance (km)
Huadian Ⅱ
2.8
Natatorium
1.0
Huadian Ⅱ
1.7
Huadian Ⅱ
1.7
Huadian Ⅱ
1.2
City Monitoring Station 1.3
Huadian Ⅱ
1.3
Huadian Ⅱ
1.0
S2 (m)
1
8720
3.5
30520
205
1
6.3 3.3
e
1
4.4 4.5
2
5110
4.0
20440
125
2
Same side
0
e
3
5300
3.0
15900
200
1
4.6 3.5
10.0 3.5
7.3 3.0
e
0
e
4
3200
3.2
3200
108
2
Opposite side
0
e
5
1690
4.8
1690
60
2
Opposite side
2
6.4 5.4
6
7
1930
6535
4.5
3.5
1930
6535
60
215
1
2
e
Same side
0
0
e
e
8
5985
3.5
5985
232
2
Same side
0
e
7.4 3.0
7.4 3.0
3.9 2.3
3.9 2.3
7.0 3.5
7.5 3.3
7.5 3.3
4.6 3.3
4.6 3.3
Abbreviation: aN- Number; bS1-Size, width height; cS2-Size, Length width (The height of the side air vent of the ventilation well is 0.4 m); dP- Position.
site with a time interval of 1 min. The measurements of traffic
volume were carried out manually by two testers standing in each
gate of the garage. They recorded the number of cars entering and
leaving the garage every 15 min during the period of measurement.
tþð
D=2
2.3. Statistical analysis
CDt ðtÞ ¼
The wastes generated from automobile were classified into
evaporative and exhaust emissions based on the types of emissions
by the U.S. Department of Transportation (2002) (Kristanto, 2006).
The amount of VOCs generated from fuel evaporation when vehicles are parked in the garage could be totally ignored because the
indoor temperature of UPGs is stabilized at less than 10 C. Thus,
only the impact of exhaust emission is focused in this study because
most pollutants (VOCs, PM, CO and NOx) are generated from the
tailpipe when the vehicle is in operation (Sawyer et al., 2000).
Therefore, the engine operating time of vehicles could directly
affect the air quality in the UPG and the hourly engine running time
is linearly correlated with traffic volume (Ho et al., 2004; Kristanto,
2006). However, the engine running time may differ from 60 to
180 s due to the different size and layout of the parking garage
according to ASHRAE (ASHRAE, 2007), therefore, the effect of the
traffic-related indoor pollutant concentration not only depends on
the traffic volumes but also depends on the size and ventilation of
the parking garage. In our study, the average ratio of occupied
parking spaces in eight UPGs is up to 92%, which means the total
number of cars in the garage is equal approximately to the total
parking space in the garage. Therefore, the average traffic activity
(ATA) was defined as the ratio of the total traffic volume (Ntot) to
total number of parking spaces (Nps) in the UPG. It is given by
equation (1):
Ntot
ATA ¼
Nps
(1)
Space utilization ratio (b) was defined as the ratio of the total
area of parking space to that of the garage, and it is given by
equation (2):
b¼
Nps A0
A
in order to analyze the correlation of indoor and outdoor pollutant
concentrations, as well as that of outdoor pollutant concentrations
and traffic volume, the measured data should be converted to the
same time interval calculated by ) (Ho et al., 2004):
(2)
A0: The average area of a parking space (2.6 5.3 m).
The time intervals of the measured data including traffic volume, indoor and outdoor pollutant concentrations were different
from each other, 15 min, 1 min and 60min, respectively. Therefore,
Cðt 0 Þdt
(3)
tD=2
The average values over time intervals of 15 min and 60 min
were denoted by C15(t) and C60(t), respectively. The impacts of
traffic volume and ambient environment on indoor air pollutants
were determined by employing Pearson correlation by SPSS 16.0.
Due to the different sources and propagation mechanisms of indoor
pollutants, their concentration is not only linearly related to one
influence factor. Therefore, a multiple regression was performed to
correlate hourly indoor pollutant concentration with the traffic
volume, ambient level and accumulative effect. The model is given
by equation (4):
C in ¼ a C in
t
t1
þ bTt þ g C out þ ε
t
(4)
where Cin is the hourly average indoor concentration, Cout is the
hourly average outdoor concentration and T is the hourly number
of cars entering or leaving the garage. The multiple linear regression model of hourly pollutant concentration in garage was also
analyzed by SPSS 16.0.
3. Results and discussion
3.1. Comparison of indoor and outdoor pollutant concentrations
The summary of the indoor thermal environment and daily
average indoor and outdoor pollutant concentrations in each
sample was listed in Table 3, and the traffic-related information was
shown in Table 4.
Compared with indoor and outdoor PM2.5 concentrations, it is
found that the indoor PM2.5 concentration is generally slightly
higher than the outdoor, with the average I/O value of 1.16. The
daily average indoor PM2.5 concentration of Case 2 was the highest
among all the samples that was 199.60 mg/m3, followed by Case 1,
Case 6, Case 8, Case 7, Case 3, Case 4 and Case 5 that was 187.90 mg/
m3, 113.73 mg/m3, 101.97 mg/m3, 88.92 mg/m3, 57.71 mg/m3, 49.66 mg/
m3 and 46.97 mg/m3, respectively. While the traffic volume of Case 1
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
629
Fig. 1. The location of sampling sites and monitoring stations in Baoding, China. Notations: the AQI data used in the above figure are derived from the real-time data (6:00 p.m. 12/
04/2018) provided by the monitoring stations during the survey period.
Table 2
The measurement range and accuracy.
Measured
parameters
Instrument
Measurement
range
Accuracy
PM2.5/PM10
CO2
Temperature
RH
TVOC
TSI 8520
TSI 7515
TSI 8392A
0.1e10 mm
0e5000 ppm
17.8e93.3 C
0e95%
0e50 ppm
± 0:1%
± 50ppm
± 0:3 C
± 3%
þ1.5%
Particle Plus 7301-AQM
was the highest that was 455 vehicles per day, followed by that
of Case 3, Case 8, Case 2, Case 4, Case 7, Case 5 and Case 6 that was
only 56 vehicles per day. In conclusion, the indoor PM2.5 concentration was mainly affected by outdoor environment rather than
traffic volume, so traffic-related emissions are not the dominant
contributors to indoor PM2.5 concentration.
However, for PM10, the daily average concentration indoors was
much higher than that outdoors, with the average I/O value of PM10
was about 4.0. The average indoor PM10 concentration of Case 3 was
the highest among all the samples, and the value was 913 mg/m3
which was 15.741 times of the outdoor (58 mg/m3). While the indoor PM10 concentrations of Case 4 and Case 5 were the lowest
which was 238.44 mg/m3 and 246.39 mg/m3, respectively, and the
outdoor concentrations were 65 mg/m3 (I/O ¼ 3.668) and 74 mg/m3
(I/O ¼ 3.330). The results indicated that the concentration of PM10
measured at the sampling site was generally far higher than the
ambient concentration. In addition, the daily average concentration
of PM10 in the garage was correlated to the total number of vehicles
entering or leaving this garage, where the Pearson correlation coefficient was 0.605 (p > 0.05). It is concluded that the traffic-related
emissions rather than ambient pollutants are the dominant contributors to the PM10 concentration of the UPGs.
630
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
Table 3
The indoor thermal environment and daily average indoor and outdoor pollutant concentrations in each sample.
Case No.
Indoor (N ¼ 1441)
Temperature ( C)
Relative humidity (%)
PM2.5 (mg/m3)
PM10 (mg/m3)
CO2 (ppm)
TVOC (ppb)
Outdoor (N ¼ 25)
PM2.5 (mg/m3)
PM10 (mg/m3)
1
2
3
4
5
6
7
8
Avea
SDb
Ave
SD
Ave
SD
Ave
SD
Ave
SD
Ave
SD
8.94
9.91
40
4.13
187.90
34.43
535.18
197.92
535
45.77
10
21.85
10.82
5.92
54
3.57
199.60
39.14
477.56
138.65
529
31.12
14
20.00
9.71
7.47
28
1.52
57.71
16.18
913.00
782.65
500
43.80
37
72.15
8.66
9.21
36
2.87
49.66
8.84
238.44
164.72
534
56.86
15
27.02
7.99
12.41
24
2.54
46.97
10.41
246.39
213.41
421
23.18
4
13.35
8.58
5.02
45
3.11
113.73
29.14
440.29
257.92
471
23.01
46
79.94
11.70
7.51
65
6.33
88.92
44.34
303.72
167.79
806
79.00
29
54.20
13.29
14.57
48
7.44
101.97
48.58
558.75
187.86
772
83.22
36
63.72
Ave
SD
Ave
SD
194
60.13
277
69.56
197
44.59
281
88.00
32
16.80
58
22.68
38
7.67
65
15.66
43
13.65
74
19.62
116
30.19
178
47.55
85
41.07
181
30.18
94
63.10
400
260.75
0.968
1.932
1.013
1.700
1.804
15.741
1.307
3.668
1.092
3.330
0.980
2.473
1.046
1.678
1.085
1.397
I/O c
PM2.5
PM10
Notations:
a
Average;
b
Standard derivation; c I/O The ratio of 24 h average indoor and outdoor pollutant concentration.
Table 4
The summary of traffic-related information in each sample.
Case No.
1
2
3
4
5
6
7
8
Total traffic volume (cars/24 h)
Average traffic activity
Space utilization ratio(%)
455
2.22
32
169
1.35
34
308
1.54
52
139
1.29
46
81
1.35
49
56
0.93
43
147
0.68
45
263
1.13
53
Among all the samples, the daily average concentration of CO2
was about 500 ppm, and that of Case 8 was the highest which was
806 ppm, followed by Case 7 of 772 ppm. Meanwhile, the daily
average TVOC concentration measured in garage was about 24 ppb.
The daily average TVOC concentration of Case 6 was the highest
which was 46 ppb, while that of Case 5 was the lowest which was
only 4 ppb. By comparing the measured data above, it was difficult
to find obvious relationship between indoor TVOC concentration
and traffic-related data. Nevertheless, Kim et al. had different discoveries in the study about VOCs of parking garages in US. Kim et al.
assessed the level of air pollution in urban parking lots and the
impact of vehicle activity on their levels. It was found that the traffic
volume on the weekend was 12 times lower than that during the
weekday, while the median concentration of air pollutants was also
reduced by 2 (CO)-7 (pPAH) times. The median values of TVOC
during weekday/weekend was 10.6/0.9 mg/m3, respectively, which
could not be directly compared with the results of this study due to
inconsistency in measurement methods (Kim et al., 2007). In
addition, Batterman et al. reported on VOC concentrations and
emissions at 15 residential garages in Michigan that varied in type,
size, use and other characteristics. Results showed that the mean
concentration of TVOC in one garage was 250.67 mg/m3, which was
far higher than the concentration measured in this study (converted to 89.7 mg/m3). It was also found that VOC concentrations
were not significantly correlated to garage characteristics or
meteorological factors, but were largely dependent on VOC sources
present and occupant activities (Batterman et al., 2006). Thus, the
daily variation characteristics of TVOC concentration in measured
UPGs was essential in the following sections.
3.2. Over standard rate
There is no specific guideline or standard on the air quality of
UPGs. Therefore, several international and Chinese standards were
combined in this study to evaluate the air quality and calculate the
over standard rate of air pollutants in garage. The daily average
concentration of PM2.5 suggested by WHO Air Quality Guideline
(AQG) was set at 25 mg/m3 (WHO, 2006; WHO, 2010), while China
has set a ceiling limit for PM2.5 concentration of 75 mg/m3 for indoor
environment in the residential area (China, M. E. P., 2012). For daily
average concentration of PM10, WHO air quality guideline (WHO,
2006) has recommended the 50 mg/m3 as a set value, as well as
EU set value (European Union, 2008). Compared with WHO,
another threshold value of PM10 concentration was set as 150 mg/
m3 according to the national standards set by both US Environmental Protection Agency (EPA) (U. S., EPA, 1990) and Chinese
government (China MEP, 2002). The indoor CO2 concentration at no
more than 700 ppm above the ambient condition (ranges from 300
to 500 ppm) would satisfy the majority of occupants suggested by
ASHERAE (2010). Therefore, it is common that 1000 ppm was set as
the threshold value for indoor CO2 concentration in buildings
(China MEP, 2002; ASHERAE, 2010). The total VOC concentration
suggested by Chinese standards was set at 0.6 mg/m3 for 8 h (China
MEP, 2002; China MOHURD, 2010). In our study, we measured the
total VOCs concentration (in ppb) in each UPG. To compare with the
standard of TVOCs, the average percentages of the main components of VOCs in parking facilities should be determined firstly, and
then convert the data from ppb units into mg/m3 units. According
to the previous studies, eight most abundant VOCs emitted in
vehicle exhaust have been identified in many parking facilities
(Batterman et al., 2006; Batterman et al., 2007; Glorennec et al.,
2008; Kristanto, 2006), and benzene, toluene, ethylbenzene and
m,p-xylenes (BTEX) and styrene are as aromatic hydrocarbons
which categorized as hazardous air pollutants (HAP) by U.S. EPA
(U.S.EPA, 1999). The average percentages of target components of
VOCs in garages presented in previous studies were shown in
Table 5. It can be concluded that the average rates of five main
components among different types of parking garages are relatively
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
631
Table 5
Average percentage of target components of VOCs in garages according to previous studies.
Type
N
Residential attached garage
Commercial underground parking facility
Residential attached garages
Residential garages(including attached, detached and carport) (in Los Angeles, CA)
(in Elizabeth, NJ)
(in Houston, TX)
Residential underground car park (level 1, working days)
(level 2, working days)
(Level 1, weekends)
(Level 2, weekends)
Residential attached garages (working days)
Residential attached garages
10
8
15
38
21
55
20
20
10
10
7
15
Reference
Kristanto (2006)
Kristanto (2006)
Batterman et al. (2006)
Hun et al., 2011
Hun et al., 2011
Hun et al., 2011
Marc et al., 2016
Marc et al., 2016
Marc et al., 2016
Marc et al., 2016
Marc et al., 2016
Batterman et al., 2007
Percentages(%)
B
T
E
X
S
8
16
11
11
8
18
17
17
20
21
15
11
58
40
58
53
54
51
41
43
37
39
49
58
9
9
6
7
8
6
7
7
8
7
8
6
24
20
25
21
22
19
23
22
23
22
20
25
1
15
0
8
7
7
12
12
12
10
9
0
Abbreviation: N: The number of samples; B-Benzene; E-Ethylbenzene; X-m,p-xylene; S-Styrene.
Table 6
Molecular weight, conversion factor and estimated percentages of target components of VOCs in parking facilities.
Pollutant
Molecular weight (Kristanto, 2006)
Conversion factor ppb to mg/m3(Kristanto, 2006)
Estimated percentage (%)
Benzene
Toluene
Ethylbenzene
Xylene
Styrene
78.1
92.1
106.2
106.2
104.2
3.194 103
3.509 103
4.349 103
4.349 103
4.225 103
13
52
8
23
4
Table 7
Comparison of target pollutants concentrations and over standard rates in each sample.
Case No.
Indoor
Mean
SD
Min
PM2.5 (mg/m ) Measuring time ¼ 24 h (0:00 a.m.-24:00 p.m.)
1
187.90
34.43
131.99
2
199.60
39.14
152.11
3
57.71
16.18
28.42
4
49.66
8.85
35.81
5
46.97
10.41
20.31
6
113.73
29.14
70.04
7
88.92
44.34
35.05
8
101.97
48.58
41.23
3
PM10 (mg/m ) Measuring time ¼ 24 h (0:00 a.m.-24:00 p.m.)a
1
535.18
197.92
292.50
2
447.56
138.65
250.10
3
913.02
782.65
50.77
4
238.44
164.72
77.23
5
246.39
213.41
91.17
6
440.29
257.92
150.06
7
303.72
167.79
127.52
8
558.75
187.86
250.05
TVOC (ppb) Measuring time ¼ 8 h (6:00 a.m.-2:00 p.m.)
1
20
33.02
0
2
21
26.32
0
3
94
102.13
0
4
16
24.45
0
5
9
19.93
0
6
63
78.33
0
7
50
73.80
0
8
74
88.90
0
CO2 (ppm) Measuring time ¼ 24 h (0:00 a.m.-24:00 p.m.)
1
535
45.77
464
2
529
31.12
495
3
500
43.80
439
4
534
56.86
440
5
421
23.18
390
6
472
23.01
420
7
806
79.00
563
8
772
83.22
650
3
Reference values
Over standard rate (%)
Max
(exposure time)
(reference values)
245.39
288.56
144.54
78.82
122.78
173.34
187.00
220.98
AQG ¼ 25 mg/m3 (24 h)
GB ¼ 75 mg/m3 (24 h)
652(AQG)
698(AQG)
131(AQG)
99(AQG)
88(AQG)
355(AQG)
256(AQG)
308(AQG)
151(GB)
166(GB)
N
N
N
52(GB)
19(GB)
36(GB)
2358.00
1346.30
4834.29
1117.20
2697.47
1894.80
849.38
1264.60
AQG ¼ 50 mg/m3 (24 h)
EPA ¼ 150 mg/m3 (24 h)
GB ¼ 150 mg/m3 (24 h)
970(AQG)
795(AQG)
1726(AQG)
377(AQG)
393(AQG)
781(AQG)
507(AQG)
1018(AQG)
257(GB)
198(GB)
509(GB)
59(GB)
64(GB)
194(GB)
102(GB)
273(GB)
209
125
425
132
145
281
437
424
GB ¼ 0.60 mg/m3 (8 h)
(approximately 160 ppb)
N
N
N
N
N
N
N
N
939
713
762
810
582
555
1142
1126
ASHRAE ¼
1000 ppm(24 h)
GB ¼ 0.10% (24 h)
N
N
N
N
N
N
N
N
a
Abbreviation: AQG: WHO Air Quality Guideline; GB: Chinese national standard; EPA: U.S. Environmental Protection Agency; ASHRAE: ASHRAE Standard 62.1, Ventilation for
Acceptable Indoor Air Quality; N: The air pollutant concentration does not exceed the reference value of the standards or guidelines.
632
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
similar. Toluene accounts for the highest proportion (37e58%),
followed by m,p-xylene(19e25%), benzene(8e21%), ethylbenzene(6e9%), and styrene(0e15%), and the molecular
weight, conversion factor and estimated percentages of target
components of VOCs were shown in Table 6. Therefore, the conversion factor of TVOCs in parking facilities was 3.757 103
(1 ppb ¼ 3.757 103 mg/m3), and the threshold of 8-h average
TVOCs concentration specified in Chinese standards (GB) was set at
approximately 160 ppb. The target pollutants concentrations and
over standard rates in all the samples were listed in Table 7. Comparison of indoor and outdoor concentrations of PM2.5, PM10 and
TVOC in each parking garage was further shown in Fig. 2, Fig. 3 and
Fig. 4, respectively.
The PM2.5 concentrations of all the samples exceeded the
limited threshold value suggested by WHO AQG and the average
over standard rate was 323%. WHO Working Group on “health aspects of air pollution”(WHO, 2004) suggested that the relative risk
percentage of PM2.5 mortality would increase 0.6% per 10 mg/m3
under the condition of continuous exposure for 1 days at the current exposure level of air pollution in North America. Pope et al.
also proposed that the relative risk percentage would increase 6%
per 10 mg/m3 when the duration of continuous exposure increased
to 16 years (Pope et al., 2002). In addition, the indoor PM2.5 concentration in five garages exceeded the standard value of the Chinese standard, the over standard rate of Case 2 was the highest
(166%), followed by that of Case 1(151%), Case 6(52%), Case 8(36%)
and Case 7(19%). The 24 h average PM10 concentrations of all the
samples were considerably higher than the air quality set value
suggested by US EPA and Chinese national standard, not to mention
the WHO AQG. The average over standard rate of PM10 measured in
all the measured UPGs was 821% based on WHO AQG, while that
was 207% based on Chinese national standard. The CO2 concentration of UPGs was generally lower than the threshold specified by
standards or guidelines. For TVOC of all the eight cases, the
compliance rate looks better than the PM mentioned above. According to the German Federal Environmental Agency (UBA)
(Umweltbundesamt, 2007), the TVOC level ranges from 0.3 to
1.0 mg/m3 when the air quality rating is good, although ventilation
may still be required to maintain a good indoor environment. In
Fig. 4, the average concentration of 8 h in each sample was within
the acceptable range, but the maximum concentrations of all the
samples were generally above the threshold value specified in GB,
with the range from 125 to 437 ppb.
Fig. 3. Comparison of indoor and outdoor PM10 concentrations in all the samples.
Fig. 4. Comparison of indoor TVOC concentration in all the samples.
3.3. Daily variation characteristics
Fig. 2. Comparison of indoor and outdoor PM2.5 concentrations in all the samples.
The daily variations of target pollutant concentrations and
traffic volume in all the samples were shown in Fig. 5.
As shown in Fig. 5, for weekdays, the number of cars entering or
leaving the garage fluctuates up and down with two distinct peaks
corresponding morning and evening rush hours in all the residential UPGs. The morning rush hour in the garage was from 7:00
a.m. to 9:00 a.m. while the evening rush hour was from 5:00 p.m. to
10:00 p.m. Among all the samples, the traffic volume always
reached its peak at about 7:30 a.m. while there is no obvious peak of
traffic volume during the evening rush hour due to the time of cars
entering or leaving the garage was relatively dispersed. Additionally, there are a small number of vehicles entering or leaving the
garage at noon (11:30 a.m.-1:30 p.m.).
The variations of the indoor PM10 concentration mainly depend
on that of traffic volume, but the variations could be also affected by
the UPGs’ structure and the existence of the air vents. For daily
variation trend, the indoor PM10 concentration of measured garage
was equal to or slightly higher than the outdoor when no car was
entering or leaving the garage, while with the growing of traffic
volume, the pollutant concentration had a considerable increase,
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
633
Fig. 5. The daily variations of target pollutant concentrations and traffic volume in all the samples.
and after the rush hour, the pollutant concentration gradually
decreased to a relatively steady value which was close to the
ambient. Therefore, it can be concluded that the variation trend of
the indoor PM10 concentration depended on that of traffic volume
to a large extent. However, the growth range of indoor PM10 concentration was not completely related to the traffic volume. For
example, although the maximum of the traffic volume in Case 1 and
Case 3 were similar, which was 23 cars per 15min and 22 cars per
15min, respectively, the maximum indoor PM10 concentration of
Case 3 (4834.29 mg/m3) was 2 times higher than that of Case 1
(2358.00 mg/m3). This difference may be due to the ventilation
effectiveness of the air vent and the different volume of the UPGs.
There was an air vent in Case 1, while no one was in Case 3.
Moreover, the volume of the garage in Case 1 was 2 times higher
than that in Case 3. Furthermore, for the Case 5 and Case 6 with the
same volume of garages, although the total traffic volume of Case 5
(81 cars per day) was higher than that of Case 6 (56 cars per day),
the daily average PM10 concentration of Case 5 was only a half of
that of Case 6. The possible explanation was the difference in the
design of air vents and entrance/exit in UPGs. For Case 5, the
634
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
Fig. 5. (continued).
entrance and exit of the garage are on opposite sides of the garage,
and the wind in the underground car park blew predominantly
from the entrance toward the exit. In addition, there are two air
vents in the garage. While for Case 6, there is only one gate for cars
entering and leaving the garage and no other air vent is designed in
this garage. Thus, the effectiveness of natural ventilation of Case 5
was relatively better. The same with variation characteristics of
PM10, the indoor TVOC concentration was completely affected by
the traffic volume. The only one difference was that the indoor
TVOC concentration was 0 when there was no car entering or
leaving the garage, which determined that the only source of TVOC
in the garage was vehicle exhaust. Furthermore, the fluctuation
ranges of the TVOC concentrations in garages were also influenced
by the traffic volume, the existence of the air vent and the structure
of the garage which was similar to the PM10. Nevertheless, unlike
PM10 and TVOC, the variation trend of the indoor PM2.5 concentration was largely consistent with the ambient PM2.5 concentration, while there were some slightly fluctuations of the indoor
PM2.5 concentration during the rush hours according to the data of
Case 3 and Case 5. In addition, the indoor CO2 concentration could
be stable in a certain range, but that would be influenced by the
traffic volume to a small extent.
3.4. Correlation analysis
The 15-min and 60-min average data of indoor pollutant
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
Table 8
Result of Pearson's correlation analysis between indoor target pollutant concentrations and traffic volume, as well as ambient environment.
Case No.
Indoor
Case 1
PM2.5
PM10
TVOC
Case 2
PM2.5
PM10
TVOC
Case 3
PM2.5
PM10
TVOC
Case 4
PM2.5
PM10
TVOC
Case 5
PM2.5
PM10
TVOC
Case 6
PM2.5
PM10
TVOC
Case 7
PM2.5
PM10
TVOC
Case 8
PM2.5
PM10
TVOC
ra
pb
R2
r
p
R2
r
p
R2
r
p
R2
R
p
R2
R
p
R2
R
p
R2
R
p
R2
R
p
R2
R
p
R2
R
p
R2
r
p
R2
R
p
R2
R
p
R2
R
p
R2
R
p
R2
R
p
R2
R
p
R2
r
p
R2
r
p
R2
r
p
R2
R
p
R2
R
p
R2
R
Traffic volume
Nc ¼ 97
Outdoor
N ¼ 25
0.423**
0.000
0.179
0.642**
0.000
0.412
0.801**
0.000
0.642
0.337**
0.001
0.114
0.333**
0.001
0.111
0.513**
0.000
0.263
0.483**
0.000
0.233
0.671**
0.000
0.450
0.695**
0.000
0.483
0.213*
0.036
0.045
0.735**
0.000
0.540
0.774**
0.000
0.599
0.074
0.470
0.005
0.463**
0.000
0.214
0.263**
0.009
0.069
0.038
0.715
0.001
0.418**
0.000
0.175
0.454**
0.000
0.206
0.127
0.215
0.016
0.780**
0.000
0.608
0.839**
0.000
0.704
0.199
0.050
0.040
0.495**
0.000
0.245
0.719**
0.945**
0.000
0.893
0.041
0.847
0.017
e
**
0.789
0.000
0.623
0.005
0.981
0.002
e
0.132
0.530
0.017
0.344
0.092
0.118
e
0.530**
0.006
0.281
0.339
0.097
0.115
e
0.956**
0.000
0.914
0.639**
0.001
0.408
e
0.178
0.394
0.032
0.240
0.249
0.058
e
0.874**
0.000
0.764
0.221
0.289
0.049
e
0.835**
0.000
0.697
0.405*
0.044
0.164
e
635
Table 8 (continued )
Case No.
Indoor
Traffic volume
Nc ¼ 97
p
R2
Outdoor
N ¼ 25
0.000
0.517
**
Correlation is significant at the 0.01 level (2-tailed).
*Correlation is significant at the 0.05 level (2-tailed).
a
Pearson's correlation coefficient.
b
Probability (2 tails significance).
c
Number of data.
concentration were calculated by equation (3) to analyze the correlation with the influence factors. The impacts of traffic volume
and ambient environment on indoor air pollutants were determined by employing Pearson correlation by SPSS 16.0, and the
result was listed in Table 8.
Apart from the data of Case 3 and Case 6, the indoor PM2.5
concentration generally exhibited a strong correlation with the
outdoor PM2.5 concentration (p 0.01) with the average R2 of
0.695, however, the indoor PM2.5 concentration of Case 3 showed a
moderate correlation with the traffic volume (p 0.01) with the R2
of 0.233, and for Case 6, the indoor PM2.5 concentration showed
inverse associations with both traffic volume and the outdoor
pollutant concentration. Five out of eight samples indicated that
the indoor TVOC concentrations were strongly correlated with
traffic volume (R2: 0.483e0.704, p 0.01), while moderate correlations were found in Case 2 and Case 6, and the weak correlation
was found in Case 5(R2 ¼ 0.069, p 0.01). For indoor PM10 concentration, all of the samples showed the existence of moderate
relationships between indoor PM10 concentration and traffic volume (R2: 0.111e0.608, p 0.01), while two of these samples also
presented the moderate correlations with the ambient environment (R2 ¼ 0.408, p 0.01 for Case 5 and R2 ¼ 0.164, p 0.05 for
Case 8). The weakly relationship between PM10 concentration and
traffic flow was also found in the results of Li et al. (Li and Xiang,
2013). One possible reason was that the indoor PM10 concentration might not only affect by traffic volume, but also associated with
the ambient and accumulative effect. Therefore, a multiple
regression was performed to correlate hourly indoor PM10 concentration with the traffic volume, ambient level and accumulative
effect, which calculated by equation (4). The multiple regression
analysis of hourly indoor PM10 concentration in each sample was
listed in Table 9.
For the Case 1, the resultant model is
C in
t
¼ 0:390 C in
t1
þ 5:363Tt þ 0:278 C out þ 150:997
t
(5)
where R is 0.872 and the adjusted R2 is 0.726. Therefore, compared
Table 9
Multiple regression analysis of hourly indoor PM10 concentration in each sample.
Case No.
Case
Case
Case
Case
Case
Case
Case
Case
1
2
3
4
5
6
7
8
Unstandardized coefficients
a
b
g
ε
0.390
0.423
0.567
0.535
0.401
0.764
0.537
0.766
5.363
4.500
26.324
11.888
17.700
34.251
10.619
4.116
0.278
0.010
0.274
1.338
2.590
0.006
0.441
0.006
150.997
247.425
81.968
42.539
102.602
32.357
156.575
91.986
R
Adjusted R2
0.872
0.572
0.933
0.978
0.895
0.944
0.973
0.953
0.726
0.230
0.853
0.950
0.772
0.877
0.940
0.895
636
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
with the consideration of a single factor, the R value for this multiple regression model could be improved by 35.8%, and the value of
R2 could be increased by 76.2%. Overall, the average R2 value of this
synthesized model was 0.780, which was 2.7 times that of the
previous model. By considering other factors affecting the indoor
PM10 concentration, this paper established a synthesized model of
relationship between indoor PM10 concentration, traffic volume,
the ambient and accumulative effect with the significant
improvement of R2.
4. Conclusions
Field measurements in eight residential UPGs under natural
ventilation were carried out in this study to identify the variation
patterns of indoor pollutant in UPGs as well as to examine the
determinants of the indoor air pollution. Based on the above
analysis, several following conclusions can be drawn:
(1) The daily average indoor concentrations of PM2.5, PM10, CO2
and TVOC in naturally ventilated UPGs were 105.81 mg/m3,
464.17 mg/m3, 571 ppm and 24 ppb, respectively. For particle
matters, the indoor PM2.5 concentrations were closely
related to that outdoors with the average I/O value of 1.6,
while the concentrations of indoor PM10 was much higher
than that outdoors with the average I/O value of 4.0.
(2) The fraction of naturally ventilated UPGs of which the PM2.5
concentration exceeded regulated air quality threshold was
100% (WHO) and 62.5 (China National Standard), and the
average over-standard rate is 323% (WHO) and 166%(China
National Standard), respectively. Meantime, the daily average
PM10 concentration of all UPGs is significantly higher than
the air quality guideline value. According to WHO AQG and
Chinese national standards, the average daily over-standard
rate was 821% and 207%, respectively. The daily average
CO2 concentration of all UPGs was not over-standard and the
8 h average TVOC concentration was lower than the
threshold specified by Chinese national standards and was
within the UBA good air quality range, but its maximum
concentration ranging from 125 to 437 ppb, were generally
above the threshold value specified by China National
Standard.
(3) The PM10 and TVOC concentrations of UPGs were moderately
correlated with traffic volume (p 0.01), and the variations
of them were also affected by the existence of air vent and
UPGs' characteristics. In comparison, the concentrations of
indoor PM2.5 exhibited a strong correlation with that of
ambient PM2.5 (p 0.01). The indoor CO2 concentration
could be stable in a certain range, though the CO2 concentration could be influenced by the traffic volume to a small
extent especially in the rush hours. Multiple linear regression
model of hourly PM10 concentration in garage including
traffic volume, the ambient and accumulative effect was
established with the average R2 value of 0.780, which was
improved by 166%.
Acknowledgment
Thanks to all the testing members who have worked tirelessly
throughout the period of measurements to bring accurate and
comprehensive data, and their names are Di Wu, Yuanwei Liu,
Haotian Niu, Qingxu Zhou, Haiyang Liu and Wenbin Zhuang from
North China Electric Power University. This work was funded by
the National Natural Science Foundation of China (No. 51708211),
the National Science and Technology Ministry of China
(2016YFC1201302), the Opening Funds of State Key Laboratory of
Building Safety and Built Environment National Engineering
Research Center of Building Technology (BSBE2017-08), Natural
Science Foundation of Hebei (No. E2017502051) and Fundamental
Research Funds for the Central Universities (No.2018MS103).
References
ASHRAE, 2007. HVAC applications. In: ASHRAE Handbook. ASHRAE, Atlanta, GA.
ASHERAE, 2010. Standard 62.1, Ventilation for Acceptable Indoor Air Quality.
American Society of Heating, Refrigerating and Air-Conditioning Engineers,
Atlanta, GA.
Batterman, S., Hatzvasilis, G., Jia, C.R., 2006. Concentrations and emissions of gasoline and other vapors from residential vehicle garages. Atmos. Environ. 40,
1828e1844.
Batterman, S., Jia, C.R., Hatzivasilis, G., 2007. Migration of volatile organic compounds from attached garages to residences: a major exposure source. Environ.
Res. 104, 224e240.
Bowatte, G., Lodge, C.J., Knibbs, L.D., Erbas, B., Perret, J.L., Jalaludin, B., Morgan, G.G.,
Bui, D.S., Giles, G.G., Hamilton, G.S., Wood-Baker, R., Thomas, P., Thompson, B.R.,
Matheson, M.C., Abramson, M.J., Walters, E.H., Dharmage, S.C., 2018. Traffic
related air pollution and development and persistence of asthma and low lung
function. Environ. Int. 113, 170e176.
Chan, M.Y., Chow, W.K., 2004. Car park ventilation system: performance evaluation.
Build. Environ. 39, 635e643.
China, MEP, 2002. GB/T188832002, Indoor Air Quality Standard. Administration of
Quality Supervision, Inspection and Quarantine, Beijing.
China, MOHURD, 2010. GB50325-2010, Code for Indoor Environmental Pollution
Control of Civil Building Engineering. Ministry of Housing and Urban-Rural
Development of the Peoples Rapublic of China, Beijing.
Costa, L.G., Cole, T.B., Coburn, J., Chang, Y.C., Dao, K., Roque, P.J., 2017. Neurotoxicity
of traffic-related air pollution. Neurotoxicology 59, 133e139.
Demir, A., 2015. Investigation of Air Quality in the Underground and Aboveground
Multi-Storey Car Parks in Terms of Exhaust Emissions. World Conference on
Technology, Innovation And Entrepreneurship, pp. 2601e2611.
European Union, 2008. Directive 2008/50/EC of the european parliament and of the
council of 21 may 2008 on ambient air quality and cleaner air for europe. Off. J.
Euro. Union 152, 16e17.
Glorennec, P., Bonvallot, N., Mandin, C., Goupil, G., Pernelet-Joly, V., Millet, M.,
Filleul, L., Le Moullec, Y., Alary, R., 2008. Is a quantitative risk assessment of air
quality in underground parking garages possible? Indoor Air 18, 283e292.
He, B.J., Zhao, D.X., Zhu, J., Darko, A., Gou, Z.H., 2018. Promoting and implementing
urban sustainability in China: an integration of sustainable initiatives at
different urban scales. Habitat Int. 82, 83e93.
Ho, J.C., Xue, H., Tay, K.L., 2004. A field study on determination of carbon monoxide
level and thermal environment in an underground car park. Build. Environ. 39,
67e75.
Hun, D.E., Corsi, R.L., Morandi, M.T., Siegel, J.A., 2011. Automobile proximity and
indoor residential concentrations of BTEX and MTBE. Build. Environ. 46, 45e53.
Jokl, M., 2000. Evaluation of indoor air quality using the decibel concept based on
carbon dioxide and TVOC. Build. Environ. 35, 677e697.
Kim, S.R., Dominici, F., Buckley, T.J., 2007. Concentrations of vehicle-related air
pollutants in an urban parking garage. Environ. Res. 105, 291e299.
Kristanto, G.A., 2006. Assessment of Volatile Organic Compounds (VOCs) in Indooor
Parking Facilities at Houston. Texas Southern University, Texas.
Li, H.R., Zeng, X.Y., Zhang, D.L., Chen, P., Yu, Z.Q., Sheng, G.Y., Fu, J.M., Peng, P.A., 2014.
Occurrence and carcinogenic potential of airborne polycyclic aromatic hydrocarbons in some large-scale enclosed/semi-enclosed vehicle parking areas.
J. Hazard Mater. 274, 279e286.
Li, Y.X., Xiang, R.B., 2013. Particulate pollution in an underground car park in
Wuhan, China. Particuology 11, 94e98.
Liu, Z.J., Cheng, K., Li, H., Cao, G., Wu, D., Shi, Y., 2018a. Exploring the potential
relationship between indoor air quality and the concentration of airborne
culturable fungi: a combined experimental and neural network modeling study.
Environ. Sci. Pollut. Control Ser. 25, 3510e3517.
Liu, Z.J., Ma, S.Y., Cao, G.Q., Meng, C., He, B.J., 2018b. Distribution characteristics,
growth, reproduction and transmission modes and control strategies for microbial contamination in HVAC systems: a literature review. Energy Build. 177,
77e95.
Marc, M., Smielowska, M., Zabiegala, B., 2016. Concentrations of monoaromatic
hydrocarbons in the air of the underground car park and individual garages
attached to residential buildings. Sci. Total Environ. 573, 767e777.
Papakonstantinou, K., Chaloulakou, A., Duci, A., Vlachakis, N., Markatos, N., 2003. Air
quality in an underground garage: computational and experimental investigation of ventilation effectiveness. Energy Build. 35, 933e940.
Pope III, C.A., Burnett, R.T., Thun, M.J., Calle, E.E., Krewski, D., Ito, K., Thurston, G.D.,
2002. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine
particulate air pollution. Jama 287, 1132e1141.
Sawyer, R.F., Harley, R.A., Cadle, S., Norbeck, J., Slott, R., Bravo, H., 2000. Mobile
sources critical review: 1998 NARSTO assessment. Atmos. Environ. 34,
2161e2181.
Song, X.C., Zhao, Y., 2017. CFD simulation of particle diffusion at the region near the
entrance or exit of an underground parking lot. 2017 2nd. International Conference on Advances on Clean Energy Research (Icacer 2017) 118, 184e188.
Z. Liu et al. / Environmental Pollution 247 (2019) 626e637
Tham, K.W., Parshetti, G.K., Balasubramanian, R., Sekhar, C., Cheong, D.K.W., 2018.
Mitigating particulate matter exposure in naturally ventilated buildings during
haze episodes. Build. Environ. 128, 96e106.
Umweltbundesamt(UBA), 2007. Beurteilung von Innenraumluftkontaminationen
mittels Referenz- und Richtwerten. Bundesgesundheitsblatt - Gesundheitsforschung -Gesundheitsschutz.
U.S., EPA, 1990. NAAQS Table. U.S. Environmental Protection Agency. Available at:
https://www.epa.gov/criteria-air-pollutants/naaqs-table.
U.S., EPA, 1999. Integrated Risk Information System (IRIS). U.S. Environmental
Protection Agency.
Vukovic, G., Urosevic, M.A., Razumenic, I., Kuzmanoski, M., Pergal, M., Skrivanj, S.,
Popovic, A., 2014. Air quality in urban parking garages (PM10, major and trace
elements, PAHs): instrumental measurements vs. active moss biomonitoring.
Atmos. Environ. 85, 31e40.
WHO, 2004. Health Aspects of Air Pollution: Results from the WHO Project" Systematic Review of Health Aspects of Air Pollution in Europe". World Health
637
Organization, Geneva.
WHO, 2006. WHO Air Quality Guidelines Global Update 2005 for Articulate Matter,
Ozone, Nitrogen Dioxide and Sulfur Dioxide. World Health Organization,
Geneva.
WHO, 2010. WHO Guidelines for Indoor Air Quality: Selected Pollutants. World
Health Organization, Geneva.
Zhang, Z.H., Khlystov, A., Norford, L.K., Tan, Z.K., Balasubramanian, R., 2017. Characterization of traffic-related ambient fine particulate matter (PM2.5) in an
Asian city: Environmental and health implications. Atmos. Environ. 161,
132e143.
Zhao, Y., Song, X.C., Wang, Y., Zhao, J.N., Zhu, K., 2017. Seasonal patterns of PM10,
PM2.5, and PM1.0 concentrations in a naturally ventilated residential underground garage. Build. Environ. 124, 294e314.
Zhao, Y., Zhao, J.N., 2016. Numerical assessment of particle dispersion and exposure
risk in an underground parking lot. Energy Build. 133, 96e103.
Thomas 1
Kaitlyn Thomas
Honors Engineering For Social Good
Research Proposal
A Detection and Avoidance System to Address Whale Strikes
Abstract
Boats passing through any area of open water are a risk to whales. Because whales are
mammals, they breathe air and in order to do that, they have to surface to the water and breathe
through their blowholes. Boats travel along the surface of the water and therefore are at risk of
hitting whales. This is a risk to both the boat and the whale.
For larger boats, it is more difficult to maneuver around a whale and so their chances of
hitting a whale in their path of transit are great. Whales being a sparsely populated marine
mammal group need to be preserved for the good of ecosystems around the globe. Boats need to
be able to spot them and maneuver around them.
By using a series of hydrophones and an AI detection system, my project hopes to solve
this problem by relaying information on the whales distance and direction from the boat to allow
the captain to maneuver around the whale and not hit it.
Intro
As boat traffic becomes more common, whale’s are at risk of being hit. Some species of
whales have become endangered due to human activity and ship striking is one of those causes.
The operation of boats is complicated on its own and spotting whales and other sea life muddles
the already difficult problems captains face on voyages. Low cost whale detection systems exist
in the research field and avoidance procedures exist in the boating world, but these two fields do
Thomas 2
not intersect. The interception of these fields will create a simpler, more effective, more
automatic whale detection and avoidance system on large boats so that they do not hit whales,
which will prevent both the damaging of boats and endangering people on the boats and the
mortality of whales due to ship strikes.
Background (/Lit Review)
Many species of whales are endangered and more are dying every year. Because of the
increase in whale beachings, which is when dead whales wash up on shores, research institutions
began conducting research to pinpoint the reason for their mortality, finding that ship-striking is
a large reason for whale mortality. (Kowalewski et. al., 2010) Large boats have a difficult time
maneuvering through the ocean at high speeds. As whales approach them, there is little to
nothing boat operators can do by the time the whale is spotted to the time the whale strikes the
boat.
Current research is focused on detecting and plotting patterns in whale’s routines as well
as their location throughout the day in an attempt to schedule boat routes around densely
populated waters as well as to better understand the nature of whales as a whole. Currently there
are two major whale detection methods being used in the research field. The first is visual. John
Durban, a Marine Mammal Biologist from the Southwest Fisheries Science Center, “[flies] a
small hexacopter to hover 100 to 200 feet above whales to take overhead photographs: from
these [he] can do photogrammetry (taking measurements from photographs) to monitor growth,
assess body condition, and identify pregnancies. These metrics allow [augmentation of]
traditional stock assessments to not only focus on abundance trends of whales, but to also
understand the underlying causes of population dynamics” (Noaa. (n.d.)). Other researchers are
Thomas 3
using satellite data to plot population density of whales. The second method is auditory. Jessica
Crance, a Research Biologist at the Alaska Fisheries Science Center, created a small sailboat
drone that is able to go into the ocean for long periods of time and record sound. She describes
her project by saying, “Our interest in the project was to determine if the Saildrone would be a
suitable platform for passive acoustic monitoring of marine mammals, in particular the critically
endangered North Pacific right whale. We attached a small, autonomous acoustic recorder called
the Acousonde to the keel of each Saildrone, and set it to record continuously up to 4 kHz, which
covers the frequency band of most marine mammals in the Bering Sea” (Noaa. (n.d.)). Google is
also conducting similar research. They use stationary hydrophones to pick up on sounds passing
them and, using an AI system, are able to detect whether or not the sound is a whale. (“Acoustic
Detection of Humpback Whales Using a Convolutional Neural Network.” (2018)). Using this
research, they have identified patterns in whale calls and in population behavior and density
throughout the waters of the pacific.
Where this research does not exist is to aid whale detection and avoidance on boats. The
research and efforts that do exist on boats have numerous problem areas. A large problem area is
the detection phase. The primary reliance for spotting whales is with human vision, usually
involving binoculars. In order for this technique to be successful, a person needs to be constantly
on the lookout for whales and looking in the right spot to detect the whale with enough time to
maneuver out of the way. A group of researchers and engineers designed a process for
maneuvering and avoidance in large boats based upon existing technology on boats. It involves a
tree of events and steps to take depending upon the circumstance (Gende, S. M. et al. (2019).
However the detection technology used in research and the processes for boat
maneuvering are not acting together.
Thomas 4
Project
Whales are dying due to ship striking, which not only is harmful to whales and the whale
population, but also to ocean ecosystems as a whole; whales need to survive. In order to achieve
this peaceful existence, boats need to employ an automatic detection and manual avoidance
system of whales so as to avoid ship strikes and decrease whale mortality therefore increasing
whale population size and preserving ocean ecosystems.
Methods and Materials
By using a similar method to the Saildrones, I will mount hydrophones onto a boat. The
microphones will pick up the sound of the whale and an AI system, similar to Google’s system
will be able to separate out the non-whale sounds from the whale sounds. After this process, the
interpretation of the sound of the whale will be able to tell the distance of the whale by relating
the decibels heard from the hydrophones in the array to the distance between the hydrophones. I
have to develop math based on the speed of sound underwater, but I can do this through real
world application using speakers and microphones. Because the hydrophones will be in a
triangular shape, this will also be able to tell general direction using similar math.
The hydrophones will pick up sound only near the surface of the water due to the
presence of potential convergence zones or sound channels blocking the sound. However, this
should not be a major problem due to the position of the boat being on the surface. A whale
cannot hit the boat from far below the boat, meaning that the sounds, even with the presence of
oddities in the water channel will still be accurately pinpointing the whale when it is in proximity
to the boat.
Thomas 5
The hydrophones will be constructed using contact microphones and an aluminum can.
Based on a design from a blogpost. (Soundfly Partners. (2017))
After this information is relayed to the captain, he or she will be able to employ a
maneuver based on the process that was outlined in the article by Scott M. Gende, Lawrence
Vose, Jeff Baken, Christine M. Gabriele, Rich Preston, and A. Noble Hendrix.
The AI system will be coded with fast.ai and their jupyter notebook through
salamander.ai. The data sets for learning and validating will be obtained through Moby Sound: A
reference archive for studying automatic recognition of marine mammal sounds. I will use the
sound files of one specific endangered whale in the Pacific Ocean: the humpback whale.
Depending on the time it takes to train the AI to sense this species and the similarities in sound
patterns across various species of whales, I may also take the time to train the AI for other
species. However I will start with the humpback.
This system will be validated by playing sounds from Moby Sound through an
underwater speaker or a speaker in a plastic bag into the pool to hydrophones on the other side
and manually feeding the sound into the Ai system. If the system sorts the sounds accurately
with the same accuracy as the trained model, the system is successful.
Aluminum Can
Contact Microphone
GPU Processor (Salamander.ai)
Arduino
Timeline
Thomas 6
Timeline
Thomas 7
Works Cited
Acoustic Detection of Humpback Whales Using a Convolutional Neural Network. (2018,
October 29). Retrieved from https://ai.googleblog.com/2018/10/acoustic-detection-ofhumpback-whales.html.
Berman Kowalewski, M., Gulland, F. M. D., Wilkin, S., Calambokidis, J. Mate, B., Cordaro, J.,
… Dover, S. (2010). Association Between Blue Whale (Balaenoptera musculus)
Mortality and Ship Strikes Along the California Coast. Aquatic Mammals, 36(1), 59–66.
doi: 10.1578/am.36.1.2010.59
Gende, S. M., Vose, L., Baken, J., Gabriele, C. M., Preston, R., & Hendrix, A. N. (2019). Active
Whale Avoidance by Large Ships: Components and Constraints of a Complementary
Approach to Reducing Ship Strike Risk. Frontiers in Marine Science, 6. doi:
10.3389/fmars.2019.00592
How does sound in air differ from sound in water? (2018, December 18). Retrieved from
https://dosits.org/science/sounds-in-the-sea/how-does-sound-in-air-differ-from-sound-inwater/.
Mellinger, D. K., & Clark, C. W. (2006). MobySound: A reference archive for studying
automatic recognition of marine mammal sounds. Applied Acoustics, 67(11-12), 1226–
1242. doi: 10.1016/j.apacoust.2006.06.002
Noaa. (n.d.). Tracking Technology: The Science of Finding Whales. Retrieved from
https://www.fisheries.noaa.gov/feature-story/tracking-technology-science-findingwhales.
Thomas 8
Siderius, M. (n.d.). High Frequency Acoustic Channel Characterization for Propagation and
Ambient Noise. Retrieved from https://www.onr.navy.mil/en/-/media/Files/AnnualReports/Ocean-Acoustics/FY07/oasider.
Siderius, M., & Porter, M. (n.d.). Effects of Sound on the Marine Environment. Retrieved from
https://www.onr.navy.mil/en/-/media/Files/Annual-Reports/MMB/FY08/mbsideri .
Soundfly Partners. (2017, October 23). How to Build a DIY Hydrophone. Retrieved from
https://flypaper.soundfly.com/produce/how-to-build-a-diy-hydrophone/.
Sue E. Moore, Randall R. Reeves, Brandon L. Southall, Timothy J. Ragen, Robert S.
Suydam, & Christopher W. Clark. (2012). A New Framework for Assessing the Effects
of Anthropogenic Sound on Marine Mammals in a Rapidly Changing Arctic. BioScience,
62(3), 289-295. doi:10.1525/bio.2012.62.3.10
US6424596B1 - Method and apparatus for reducing noise from near ocean surface sources.
(n.d.). Retrieved from https://patents.google.com/patent/US6424596B1/en.
Purchase answer to see full
attachment