You are on page 1of 11

KSCE Journal of Civil Engineering (2018) 22(1):330-340 Transportation Engineering

Copyright ⓒ2018 Korean Society of Civil Engineers


DOI 10.1007/s12205-017-1538-1 pISSN 1226-7988, eISSN 1976-3808
www.springer.com/12205
TECHNICAL NOTE

Level of Service for On-street Parking


Debasish Das* and Mokaddes Ali Ahmed**
Received September 5, 2016/Revised November 21, 2016/Accepted January 18, 2017/Published Online March 22, 2017

··································································································································································································································

Abstract

Parking of vehicle is one of the most important issues of the urban transportation system in any Central Business District (CBD). A
sharp rise in vehicular ownership and insufficient transit system give rise to the huge on-street parking demand leading to a reduction
in effective carriageway width, flow speed, creating unnecessary congestion, etc. The problems can be controlled by providing an
efficient parking system. The present study aims to estimate the Level of Service (LOS) of on-street parking. LOS is calculated based
on three parking selection criteria, viz. Parking Characteristic (PC), Safety Characteristic (SC) and Design Characteristic (DC). Four
CBDs of Kolkata Metropolitan Area (KMA) are selected as a case study area. The overall LOS equation is formulated by Analytic
Hierarchy Process (AHP). Cluster analysis and Zero dimension process are used to standardize the variables to designated LOS.
Various surveys are carried out for data collection. The data are analyzed using SPSS. Finally, the overall LOS for on-street parking is
estimated for all the selected CBDs.
Keywords: CBD, parking facility, AHP, Clustering analysis, Zero dimension process, LOS estimation
··································································································································································································································

1. Introduction Keese (1964) described the level of service concept based on


demand and capacity for highway, and also discusses various
Due to the increase of the personalized vehicle and lack of variables which affect the level of service for traffic engineering
sufficient parking spaces, the parking has become a major as well as for pavements. Another study (Yu et al., 1973) established
problem in the urban transportation system. Mainly on-street a method to find the individual parking facility efficiency based
parking is observed in KMA area. The users are forced to park on the LOS concept. Their objective was to find the principal
their vehicles on-street due to the lack of off-street parking, and variables which are significant for parking service and identifies
this creates cruise of parking (Shoup, 2007).Cruising of parking that parking availability and flow/capacity ratio as the important
also has an adverse impact on the Environment. It is essential to variable. The concept of AHP was used for analytical planning
modify the operation system for parking based on user’s demand in later stage (Saaty et al., 1985). Parking standards developed
and space availability. To modify an existing parking system, it is based on LOS idea for structured parking (Smith, 1985). Smith
important to understand the service quality of the on-street (1996) defined the Crime Prevention through environmental
parking i.e. the LOS of the on-street parking. design in parking facilities and LOS approach based on safety
LOS concept is widely used for roadway safety and design, but and security. The author also described various safety and
few works have been found applying the concept in the field of security tools like Lighting, Natural surveillance, Access control,
on-street parking. The aim of the study is to evaluate the LOS of Stair towers and elevators, Panic buttons and emergency phones,
on-street parking. Sound surveillance, CCTV, Security personnel etc. for design a
parking lot. The effect of interventionist parking policy, high
2. Literature Review parking price, limited parking supply, insufficient transit system
and a modal shift from private vehicles to transit ridership, on the
The importance of parking LOS as a component in the urban CBDs’ parking system have been discussed in a study (Mildner
transportation system has attracted many studies in the past. The et al., 1997). These parameters are also relevant while estimating
literature ranging more than four decades have been studied. The the on-street parking LOS.
literature divided into three paragraphs based on the year of In 2000 another level of service approach was developed by
publication i.e. from 1965-1999, 2000-2010 and 2011-2016. The Key et al. (2000). They interpreted LOS based on flow condition
detailed on the studies have been discussed below in subsequent and recommend four acceptable level of service A B C and D for
paragraphs. parking services from the best to the worst situation. Based on

*Ph.D. Scholar, Civil Engineering Dept., NIT Silchar, Assam 788010, India (Corresponding Author: E-mail: debasishd89@gmail.com)
**Professor, Civil Engineering Dept., NIT Silchar, Assam 788010, India (E-mail: ali.mokaddes@gmail.com)

− 330 −
Level of Service for On-street Parking

LOS approach Smith recommends design criteria for structure al., 2013). By traffic conflict analysis technique and traffic
parking. A parking policy evaluation model was developed by survey, they describe the relationship between events of the
Shiftan et al. (2000). Very few works on parking studies have conflicts and the velocity of parked vehicles and non-motorized
been conducted in China. The fuzzy model used for performance vehicles, bicycles, and e-bike volume, the width of the bicycle
analysis of parking facility (Dong et al., 2009). Three aspects, lane, which designate the level of service of the curbside parking
namely safety, convenience and efficiency, are considered in facility. Smith (Smith, 2013) provides evidence that parking
their work. But for estimating the service quality of the parking demand is reduced along the light rail transit line. The minimum
they did not perform any survey and analysis. The proposed parking demand in off-season also estimated from this study. The
methodology was primarily based on experts’ objective evaluation. result of parking demand helps the planner to construct or
Kononov et al. (2003) found out the LOS of safety for highway improve any parking lot to maximize the efficiency of that lot.
using safety performance function. The analysis was done to Some particular area of concerns has been identified for
estimate total accident or fatality in roads. The LOS concept in evaluating the parking problems in Kolkata. The condition of the
roads is a quite known process to calculate the accident statistics parking of the Kolkata also discussed is a study (Remo et al.,
and safety issues of roads. The idea of this study can be partially 2015). The performance on-street parking estimated based on
considered for calculating the same while estimating the on- automated parking transaction data (Cats et al., 2016). They
street parking LOS. The walking distance to the destination is the introduce a survey methodology which measures the impacts of
primary criteria for parking facility‘s performance evaluation, on-street parking. The survey method compares variables, including
i.e. the LOS (Smith et al., 2008). They have set some standard their temporary variables. Shaaban et al. (2016) carried out a
walking distance for a different LOS. The parking service level study on factors affecting parking choice in Qatar. The study
can be estimated by using the multivariate logit model (Yang et aims to find out the most significant parameters for selecting a
al., 2010). The analysis used users’ survey data, but the result parking area using classification tree analysis. Four parking areas
gives a large gap compared to real world situation. A report by have been considered, and the study shown walking distance
Kolkata Municipal Corporation (2000) described the parking from the parking area to destination plays the most significant
condition and parking policy for Kolkata. The study indicates role in selecting a parking area.
some parameters on on-street parking areas under KMA like, on- Very few works have been done till now worldwide in the field
street parking locations, parking fee, parking duration, the of parking LOS estimation. This study aims to estimate LOS of
operational system of the parking areas which help to conduct the on-street parking by in incorporating some new variable like
the present study. the distance between two consecutive parked vehicles (AS). The
Victoria Transport Policy Institute (2011) defined LOS by detailed estimation methodology to find out LOS in discussed in
taking the walking distance to the destination as the primary the next section.
criteria for parking facilities performance evaluation. They have
set some standard walking distance for different LOS. K-means 3. Methodology
clustering analysis defined the four levels of service criteria,
taking four service variables, i.e. the ratio of peak-hour demand The overall LOS equation is developed using users’ perception
to capacity, the average parking space occupying rate, parking data. AHP process is used to estimate the co-efficient of PC, SC,
cost, and circulation time as the most effective parameters (He et and DC and formulate an overall LOS equation. AHP is a
al., 2012). On-street parking level of service also evaluated based decision-making tool to find the weight of different elements
on parking inventory and land use (Moeinaddinia et al., 2013). hierarchically. The responses for selecting a parking area based
Based on space utilization, no. of legal and illegal parking, on PC, SC, and DC are collected from a questionnaire survey.
parking shades, accessible walking routes and accessible transit Selecting a parking area based on PC, SC and DC priority wise
station they evaluate the parking areas level of service. The LOS varies from respondents to respondents, which is shown in Table
of curb parking estimated based on parking accessibility (Ye et 4. The coefficient value for PC, SC and DC are found out in five

Table 1. Calculation of Priority Vector


Preference Preference weightage Priority vector

1st preference (P1) Pw1 ∑ n( Preference weightage of P1 )


Pw1 = --------------------------------------------------------------------------------------
n

2nd preference (P2) Pw2 ∑ n( Preference weightage of P2 )


Pw2 = --------------------------------------------------------------------------------------
n

3rd preference (P3) Pw3 ∑ n( Preference weightage of P3 )


Pw3 = --------------------------------------------------------------------------------------
n
Note: n= total number of respondents

Vol. 22, No. 1 / January 2018 − 331 −


Debasish Das and Mokaddes Ali Ahmed

steps using AHP as discussed below: against each preference, and the result is shown in Table 7.
Step I: The first step is to evaluate the priority vector of the
ai j
preferences indicated by the users. The users are also asked to xij = -----------------
3
(5)
rank their preferences. The priority vector for a particular ∑ i = 1aij
preference is evaluated by taking the average of priority of all the Where, i = parking selection criteria
data which is shown in Table 1. The result is shown in Table 5. j = Parking preference
Step II: The second step is to create the comparison matrix xij = (i,j)th element of normalized matrix
[Ai] for each preference from Eq. (1). The result is indicated in aij = (i, j)th element of Eigenvector
Table 6. The questionnaire survey record preference wise For example: the elements of the normalized matrix for P1, i.e.
selecting a parking are by the respondents. j = 1 are obtained using the following steps
a
11
wi 1 w i 1 w i 1 When, i = 1, x11 = -----------------------------
-
------- ------- ------- a11 + a21 + a31
wi 1 w i 2 w i 3
a
w
Ai = ------i-2 w w
------i-2 ------i-2 (1) 21
When, i = 2, x21 = -----------------------------
-
wi 1 w i 2 w i 3 a11 + a21 + a31
wi 3 w i 3 w i 3
------- ------- ------- a
wi 1 w i 2 w i 3 31
When, i = 3, x31 = -----------------------------
-
a11 + a21 + a31
Where, i = preference sequence
When, Similarly, the element of the normalized matrix for P2 (j = 2)
i = 1, i.e. at P1 and P3 (j = 3) also estimated and shown in Table 8.
W11 = Frequency of choosing PC as first preference Consistence Index (CI) and Consistence Ratio (CR) are calculated
W12 = Frequency of choosing SC as first preference to validate the AHP result. CI for each variable and overall
W13 = Frequency of choosing DC as first preference hierarchy can be obtained by hand calculation. Initially, each
Similarly, column of the judgment matrix is added. Then multiplication of
i = 2, the sum of the first column with the value of the first component
W21 = Frequency of choosing PC as second preference of the normalized priority vector, the sum of the second column
W22 = Frequency of choosing SC as second preference by that of the second component and so on. CI is divided by the
W23 = Frequency of choosing DC as second preference Random Consistence (RC) to obtain CR. The CR value should
i= 3, be around 10% or less to be acceptable. But in some cases the
W31 = Frequency of choosing PC as third preference CR value 20% may also be acceptable, but above 20% is not
W32 = Frequency of choosing SC as third preference acceptable. In this case, the calculation needs to be revised. The
W33 = Frequency of choosing DC as third preference result is shown in Table 9.
Step III: The third step is to calculate the Eigenvector [ Xik ] Step V: the global priority matrix is obtained from the matrix
component for each parking preference by the following multiplication of the normalized matrix in step IV for each
equations. Eq. (2), Eq. (3) and Eq. (4) are used to find out the facility and priority vector matrix in step I. The result is indicated
Eigenvector for P1, P2 and P3 respectively, and the result is shown in Section 7.
in Table 7. The global priority matrix is the values for the coefficients PC,
SC, and DC. The overall LOS is computed using Eq. (6).
w w w
ai 1 = 3 ------i-1 × ------i-1 × ------i-1 (2) Variable selection: The variables, like volume to capacity ratio
w i 1 wi 2 w i 3
(V/C ratio), Average Parking Space Utilization Rate (APSUR),
w w w Average Parking Fee (APF), Average Ease Time (AET), Parking
ai 2 = 3 ------i-2 × ------i-2 × ------i-2 (3)
w i 2 wi 2 w i 3 Turnover Rate (PTR); Number of Vehicles Per Fee Collector
(VPFC), availability of lighting (L); Average Spacing (AS) are
w w w
ai 3 = 3 ------i-3 × ------i-3 × ------i-3 (4) considered to estimate PC, SC and DC respectively. For on-street
w i 1 wi 2 w i 3
parking, quality of the parking facility largely depends on space
Where, i = parking criteria availability, the capacity of the parking area and the number of
When, i =1, indicates PC fee collector. It is observed that the distance between two
And similarly i = 2, indicates SC and i = 3, indicates DC consecutive parked vehicles is not maintained in the study area.
Where, Khanna and Justo (1991) recommended AS of 0.9 m. Due to
ai1 = Eigenvector of P1, ai2 = Eigenvector of P2, ai3 = Eigenvector lack of space, AS sometimes reduces even up to 0.05 meter. The
of P3 state government has assigned some private organizations to
Step IV: The fourth step is to compute the normalized matrix manage the on-street parking in KMA. These organizations
for each preference from Eq. (5). The normalized matrix is formed appointed fee collectors to collect the revenue for parking. As
to estimate the individual weightage value for each parking criteria most of the parking areas are parallel parking, so distance

− 332 − KSCE Journal of Civil Engineering


Level of Service for On-street Parking

Table 2. Selected Variables for Each Parking Criteria


Parking selection criteria Variables Unit
Volume to capacity ratio (V/C ratio) Unit less
Average parking space utilization rate (APSUR) Unit less
PC Average parking fee (APF) INR
Average ease time (AET) Minute
Parking turnover rate (PTR) Vehicle/hour/bay
Number of vehicles per fee collector (VPFC) Vehicle per fee collector
SC
Availability of lighting (L) LUX
DC Average spacing (AS) Meter

covered by per fee collector is in terms of no. of vehicles. In vehicle and also reduces the LOS of the parking area. So this
actual scenario, the fee collectors operate the parking area and parameter is taken as safety parameter. Parking fee is more or
instruct the drivers to park their vehicles. The fee collectors try to less fixed throughout the day in the study area. But in some
increase the no. of vehicle parked in their designated area by locations, it varies during peak demand hours and festival times.
reducing the distance between two successive vehicles to The summary of the selected variables is indicated in Table 2.
generate more revenue. Many times it is observed that the fee The data of each variable selected under each parking
collectors used to adjust the parked vehicle manually to characteristics (Table 2) are grouped into four categories/ranges.
accommodate more vehicles. While taking out the vehicles, This is done by K-Means Clustering Analysis in SPSS. The
sometimes it is observed that the vehicle used to collide with the boundary values that are obtained from the cluster analysis are
front or the rear vehicles. This decreases the safety of the parked used to establish the range of each LOS category. These clustered

Fig. 1. Methodology Flow Chart

Vol. 22, No. 1 / January 2018 − 333 −


Debasish Das and Mokaddes Ali Ahmed

Fig. 2. Survey Locations and Study Area

values are then converted to dimensionless values (described in questionnaire surveys are carried out at each location to collect
section 7.3). The mean of the dimensionless values of the the data of selected variables. The following survey techniques
variables for each parking characteristics are used as PC value, have been conducted for data collection.
SC value and the DC value in the AHP equation (Eq. (6)). The parking inventory survey: It is a method to study the
Finally, the overall LOS of the on-street parking is estimated. existing scenario of the on-street parking viz. parking facility, the
Various types of survey viz. parking inventory survey, In-out capacity of the area, etc. Distometer is used to estimate the
survey, License plate survey and questionnaire surveys are existing parking area in the parking lot.
conducted for data collection of the above mentioned variables. In-out survey: This technique is used to determine the
The detailed of survey techniques are discussed in Section 6. accumulation and the occupancy of the parking area. This survey
is used to determine the occupancy of the parking lot. Initial
5. Study Area Selection occupancy of the parking lot is taken. The number of vehicles
entering and leaving for a particular time interval is counted.
Four locations in the KMA are chosen as the case study areas Finally, occupancy of that lot is also taken. The labour required
on the basis of the intensity and type of land use and on-street for this survey is very less. Only one person is enough to conduct
parking scenario. The CBD in Kolkata is distributed over a large this survey.
space and not concentrated in a particular location like other Initially, a parking area is chosen. The area is divided into
cities. Keeping the fact in mind, on-street parking is considered subsequent sections for easiness in conducting the survey. The
based on work trips and shopping trips. Four major CBDs from parking bays are marked as 1, 2, 3,…., n, for each section. After
KMA are selected based on trip purpose. Dalhousie is primarily setting up, the survey is conducted both manually and video
office areas, Gariahat is shopping area and Park Street and graphically. The data obtained are transferred to Excel sheet and
Camac Street are a combination of work and shopping trips. The the parking accumulation and occupancy are obtained from the
locations of the CBDs are indicated in Fig. 2. excel sheet.
License plate survey: Parking duration, parking turnover, and
6. Survey Techniques parking fee are estimated from this survey. Each parking bay is
to be monitored at a continuous interval of 15 minutes, and the
Four types of surveys were carried out in this study. Parking license plate number is noted. If the time interval is minimized,
inventory survey, in-out survey, license plate survey and the chances of missing short-term parkers will be lessened. But

− 334 − KSCE Journal of Civil Engineering


Level of Service for On-street Parking

Table 3. Types of Surveys for Data Collection of Selected Variables


Sl. No Name of the survey Data collected for Variables
1 Parking inventory survey Number of vehicles per fee collector (VPFC), Availability of lightning (L)
2 In-out survey Volume to capacity ratio (V/C ratio), Parking turnover rate (PTR), Average spacing (AS)
3 License plate survey Average parking space utilization rate (APSUR)
4 Questionnaire survey Average parking fee (APF), Average ease time (AET)

Table 4. Number of Respondents’ Response for Selecting a Park- Table 5. Result of Priority Vector
ing Lot Preference weightage Priority vector
Parking PC SC DC Pw1 0.75
P1 285(W11) 91(W12) 24(W13) Pw2 0.19
P2 83(W21) 168(W22) 149(W23) Pw3 0.06
P3 32(W31) 141(W32) 227(W33)
Note: 285, 91, 24 are assigned as W11,W12, W13 respectively, and so on
for easy to understand the calculations in later stages.
and DC. The result of priority vector is shown below in Table 5.
It is observed from Table 5 that the total weightage to first
this method is very labor intensive. The parking duration for a preference is 75 percent, i.e the weightage given by the
particular vehicle in a bay is estimated. The parking fee is respondents for P1 while selecting a parking area on the basis of
estimated based on the parking duration. PC, SC, and DC is 75 percent. Similarly, it is 19 percent and 6
Questionnaire survey: The questionnaire survey is conducted percent for P2 and P3 respectively.
in all the survey locations. A qualitative and quantitative analysis Step II: The computed comparison matrix is shown in Table 6.
is carried out to understand the situation of the existing parking It is observed from Table 6 that the users prefer the PC,
and mode choice. Willingness to pay survey is also conducted to approximately 3.1 and 11.87 times more than SC and DC
understand the users’ requirement for mode shift from private respectively as their first preference. It is found from Table 6 that
vehicle to public transit. The questionnaire consists of 4 major for P2, SC is more significant than PC and DC. Preference of SC
groups such as Personal Details, Trip Characteristics, Parking is 2.02 and 1.13 times more than that of the PC and DC
Characteristics and Parking Location Choice. The data are respectively.
arranged on Excel and the analyses are carried out by using Step III: Eigenvector for each PC, SC and DC are calculated
SPSS. against each parking preference. The calculated result is shown
Survey method(s) for collecting data for above mentioned in Table 7.
variables are shown in Table 3. Step four: The elements of the normalized matrix for each
The questionnaire survey is conducted to get a glimpse of preference is estimated from Eq. (5) and the result are shown
parkers’ characteristics for choosing a parking area on the basis below in Table 8.
of PC, SC, and DC (Section 4). Collected data from all locations It is observed from Table 8, a weightage of 0.7125 is given to
are grouped together. The respondents’ response is recorded and PC against P1, where a weightage of 0.2075 and 0.0800 is given
presented in Table 4. for P2 and P3. Similarly, the weightage value is also found for SC
and DC for respective P1, P2 and P3.
7. Result and Discussion CR is needed to be estimated and validate to justify the
estimation from AHP analysis. The result is shown below.
The estimated results from AHP, Clustering analysis, and zero Where,
dimension process are shown in this section. lmax = largest Eigen value
Result from AHP analysis: Stepwise (Section 4) result from n = no of elements being compared (here, n = 3, i.e. PC, SC
AHP analysis is shown below. and DC)
Step I: The weightage given by the respondents to their CI = Consistency Index
preference for selecting a parking area on the basis of PC, SC CR = Consistency Ratio

Table 6. Comparison Matrix for Each Parking Selected Criteria Against Each Preference
Parking selection P1 P2 P3
criteria PC SC DC PC SC DC PC SC DC
PC 1 3.1319 11.875 1 0.494 0.55705 1 0.227 0.14097
SC 0.3193 1 3.79167 2.0241 1 1.12752 4.4063 1 0.62115
DC 0.0842 0.2637 1 1.7952 0.8869 1 7.0938 1.6099 1

Vol. 22, No. 1 / January 2018 − 335 −


Debasish Das and Mokaddes Ali Ahmed

Table 7. Calculated Eigen Vectors Table 9. Validation by CI and CR Estimation


Parking selection Parking preference CI
Parking λmax – n CR = ------- in
criteria P1 P2 P3 λmax cI = ----------------
- RC
PC 3.4260 0.7520 0.1585
preference n–1
percentage
SC 1.0107 1.0170 2.3643 P1 3.0007 0.00035 0.06
DC 0.2888 1.3075 2.6670 P2 3.0002 0.00010 0.017
Table 8. Result of Normalized Matrix Calculation P3 3.0005 0.00025 0.043

Parking selection Parking preference


criteria P1 P2 P3
0.06 percent, 0.017 percent, and 0.043 percent respectively. As
PC 0.7125 0.2075 0.0800
the value of CR is less than 10 percent, this result is acceptable.
SC 0.2275 0.4200 0.3525
Step five:
DC 0.0600 0.3725 0.5675
0.7125 0.2075 0.0800 0.75 0.58
Y = 0.2275 0.4200 0.3525 × 0.19 = 0.27
RC = Random Consistency = 0.58 for n= 3
It is found from Table 9 that the CR value for P1, P2 and P3 is 0.0600 0.3725 0.5675 0.06 0.15

Table 10. Result of Clustering Analysis


Variable Unit Cluster center Boundary value
V/C ratio Unit less 0.40, 0.77, 1.07, 1.23 0.59, .92, 1.15
APSUR Unit less 0.29, 0.52, 0.71, 0.92 0.41, 0.62, 0.82
APF INR 20,55,89,121 38,72,105
AET Minute 2, 4, 6 10 3, 5, 8
PTR Vehicle/hour/bay 1.32,0.84, 0.58, 0.36 1.08,0.71,0.47
VPFC Vehicle/person 6,10,14,18 8,12,16
L Lux 3.87,12.23,27.55,38.33 8.05,19.89,32.94
AS Meter 0.16, 0.54, 1.06, 1.70 0.35, 0.80,1.38

Table 11. Results of K-S Analysis


Variables Classification K-S Z value Gradual significance both ends Test of difference
T1, T2 5.252 0.000 Significant
V/C ratio T2,T3 6.838 0.000 Significant
T3,T4 8.888 0.000 Significant
T1, T2 3.922 0.000 Significant
APSUR T2,T3 4.602 0.000 Significant
T3,T4 6.162 0.000 Significant
T1, T2 7.242 0.000 Significant
APF T2,T3 5.577 0.000 Significant
T3,T4 4.550 0.000 Significant
T1, T2 6.047 0.000 Significant
AET T2,T3 4.953 0.000 Significant
T3,T4 3.862 0.000 Significant
T1, T2 4.705 0.000 Significant
PTR T2,T3 8.776 0.000 Significant
T3,T4 8.615 0.000 Significant
T1, T2 3.606 0.000 Significant
VPFC T2,T3 3.950 0.000 Significant
T3,T4 4.721 0.000 Significant
T1,T2 8.832 0.000 Significant
L T2,T3 5.797 0.000 Significant
T3,T4 4.671 0.000 Significant
T1, T2 4.392 0.000 Significant
AS T2,T3 5.932 0.000 Significant
T3,T4 8.643 0.000 Significant

− 336 − KSCE Journal of Civil Engineering


Level of Service for On-street Parking

Table 12. Classification of LOS Based on Clustering Analysis


LOS classification
Parking selection criteria Variables
LOSA LOS B LOS C LOS D
V/C ratio 0-0.59 0.59-0.92 0.92-1.15 1.15-2
APSUR 0-0.41 0.41-0.62 0.62-0.82 0.82-1
PC APF 0-38 38-72 72-105 105-240
AET 0-3 3-5 5-8 8-20
PTR 3-1.08 1.08-0.71 0.71-0.47 0.47-0
VPFC 0-8 8-12 12-16 16-30
SC
L 100-32.94 32.94-19.89 19.89-8.05 8.05-0
DC AS 3-1.38 1.38-0.80 0.80-0.35 0.35-0

Table 13. Variable Zero Dimensionless Process Based on LOS Classification


LOS category A B C D
Dimensionless Max Min Max Min Max Min Max Min
Value 1.00 0.75 0.75 0.50 0.50 0.25 0.25 0
V/C ratio* 0 0.59 0.59 0.92 0.92 1.15 1.15 2
APSUR* 0 0.41 0.41 0.62 0.62 0.82 0.82 1
APF* 0 38 38 72 72 105 105 240
AET* 0 3 3 5 5 8 8 20
VPFC* 0 8 8 12 12 16 16 30
PTR** 3 1.08 1.08 0.71 0.71 0.47 .47 0
L** 100 32.94 32.94 19.89 19.89 8.05 8.05 0
AS** 3 1.38 1.38 0.80 0.80 0.35 0.35 0
Note: * = Smaller the value, better the LOS and ** = Larger the value, better the LOS

LOS of parking area = independence of each cluster class. The result of the K-S analysis
is indicated in Table 11.
( 0.58 × PC ) + ( 0.27 × SC ) + ( 0.15 × DC ) (6)
Based on the cluster analysis result, the designated range of
Result from Clustering analysis: Variables like, V/C ratio, each variable at each LOS level is set up. The result is shown in
APSUR, APF, AET, PTR, VPFC, L and S are considered in Table 12.
evaluating LOS of on-street parking. Various surveys are conducted It is observed from above table that, if the APF is less than INR
for collecting data for these variables (Table 3). The data collected 38 then it falls in LOS A category. If the APF is in the range of
from all locations are clubbed together and K-Means Clustering INR 38 to INR 72, INR 72 to INR 105 and INR 105 to INR 240,
analysis is carried out with SPSS to standardize the range of the then it situated in LOS B, LOS C, and LOS D respectively.
individual variables. K-Means Clustering analysis is a process to Similarly, the range wise categorical class is set up in Table 12
classify data in some groups based on their nearest mean. The for each variable.
variables are classified in four groups viz. LOS A, LOS B, LOS
C and LOS D. Result of clustering analysis is given in Table 10. 7.2.1 LOS Classification and Zero Dimensional Process
Table 10 gives a result of clustering centers and boundary Four LOS categories are assigned a range of values between 0
value of individual variables. Clustering centers are the mean to 1 equally subdivided with an interval of 0.25 i.e., LOS A is
value of individual cluster, i.e. for this study. LOS A, LOS B, assigned a range of 1 to 0.75, LOS B has a range of 0.75 to 0.5
LOS C and LOS D are designated as excellent, good, fair and and so on which is termed as a zero dimension process. The
poor. For example, the collected data to estimate the cluster result is indicated in Table 13 (shown in the shaded portion). The
centers for APF, four clusters are generated based on input data variables are converted to the unit less value using this process
for the different cluster. INR 20, INR 55, INR 89 and INR 121 and the clustering values for each variable are computed using
are four cluster centers for LOS A, LOS B, LOS C and LOS D the interpolation method (Eq. (7)). The interpolation method is
respectively. From this cluster center, three boundary values are used for a detailed estimation of each CBD.
also generated, i.e. INR 38, INR 72 and INR 105. The subsequent
( y 2 – y1 )
analyses have been conducted based on these boundary values. y = y1 + -----------------
- ( x – x1 ) (7)
( x 2 – x1 )
The values are also used to standardize the APF for LOS
category. Where,
After finding out the cluster centers and cluster boundary, K-S y = Computed dimensionless value
(Kolmogorov-Smirnov) analysis is carried out to find out the y1 = Minimum dimensionless value assigned for each LOS

Vol. 22, No. 1 / January 2018 − 337 −


Debasish Das and Mokaddes Ali Ahmed

category values are then compared with the standardize values (Table 12
y2 = Maximum dimensionless value assigned for each LOS and Table 13) to determine the overall LOS. The LOS levels are
category varied from time to time of a day. The individual LOS for each
x1 = Lower limit of individual variable for each LOS category location is estimated at various times i.e. morning time (07:00 to
x2 = Upper limit of individual variable for each LOS category 12:00), afternoon time (12:00 to 18:00) and evening time (18:00
x = Estimated data for each variable from respective survey to 20:00) and indicated in Table 15.
The LOS category for the parking area in the Dalhousie area in
Table 13 shows that, the variables are converted into a fixed
the morning is B. As the office time starts, the parking demand
dimensionless maximum and minimum value for a particular
increases and the LOS reduce to category C and remain
level of service. For an example, if the V/ C ratio is 0.92, than the
unchanged throughout the afternoon. In the evening time when
dimensionless value will be 0.50, and if V/C ratio is 1.15 the
the office gets over, the LOS of the area again returns to B
dimensionless value is 0.25. Any value (for V/C ratio) lies
category. In Gariahat area, the maximum users are visiting this
between 0.92 and 1.15 V/C ratio, the dimensionless value is
area in the afternoon brings down the LOS of the parking area
calculated by interpolation method (Eq. (7)). Suppose, the V/C
from B to C category. It again returns to LOS B after 19:00
ratio is 1, the dimensionless value will be 0.413. Similarly, all the
hours. In the Park Street area, some schools are situated. All
clustering values for each variable are converted into dimension
schools are starting more or less the same time in the morning
less value. These results are used to determine the individual
time. In the morning time, due to less traffic, the LOS is found to
LOS for PC, SC, and DC.
be of A category. LOS B is reduced to LOS C in the afternoon
time due to the high demand of closing of schools and starting of
7.2.2 Average Value
the shopping season. The Camac Street is a combination of
Average of the dimensionless value of the variables for each
shopping and work trips. Though very few offices are located at
parking criteria is estimated to find out PC, SC, and DC which
this location, it still creates a huge parking demand in this CBD
are shown in Table 14.
area due to the insufficient transit system. Though LOS B is
The estimated average values of PC, SC and DC from the
observed in the morning time, but later, due to the closing of the
above table are used in Eq. (6) to get the overall LOS.
offices and starting of shopping hours, the LOS C is observed for
After setting up the standard value for each variable, the data
the rest of the day. A detailed calculation of LOS estimation of
set for individual location are estimated. The individual location
Dalhousie area is shown in Appendix A. Similarly, the LOS for
other CBDs is also estimated.
Table 14. Average Value Calculation
Parking selection 8. Conclusions
Average value
criteria
v A comprehensive parking LOS method is developed in this
-- ratio + APSUR + APF + AET + PTR
PC c study. The LOS is estimated by combining three parking
PC = -----------------------------------------------------------------------------------------
5 selection criteria, namely PC, SC and DC. Important variables
VPFC + L for each selection criteria are selected and evaluated to compare
SC SC = ------------------------
2 with the standardize table (Table 12 and Table 13). Four LOS
DC DC = AS categories viz. LOS A (excellent), LOS B (good), LOS C (fair)
and LOS D (poor) are defined based on the index value of each
Table 15. Overall Estimated LOS for All Selected Location’s Time variable, which are estimated based on the users’ perception
Wise data. AHP, clustering analysis, and zero dimension process are
Estimated overall tried to standardize the value for each variable. It is found from
Locations Time LOS value
LOS the study that the LOS is changing from CBD to CBD and from
07:00 to 12:00 0.63 LOS B time to time. The LOS of the Dalhousie is in the category of B in
Dalhousie 12:00 to 18:00 0.39 LOS C the morning and evening time. Whereas it is reduces to category
18:00 to 20:00 0.51 LOS B C in the afternoon time. Similarly, the LOS for others CBDs also
07:00 to 12:00 0.59 LOS B estimated on different time. The LOS estimation techniques used
Gariahat 12:00 to 18:00 0.44 LOS C in the study can be effectively used by the policy maker for
18:00 to 20:00 0.62 LOS B estimating the existing on-street parking LOS and accordingly
07:00 to 12:00 0.56 LOS B appropriate measures may be adopted for smooth service. The
Park street 12:00 to 18:00 0.37 LOS C on-street parking LOS varies with respect to location, time and
18:00 to 20:00 0.56 LOS B overall parking policy adopted for an urban CBD. The standard
07:00 to 12:00 0.57 LOS B AS is 0.9 m. But, due to lack of space, the calculated LOS is
Camac street 12:00 to 18:00 0.40 LOS C reduced which is the limitation of this study area. Some new
18:00 to 20:00 0.42 LOS C variables like, climatic condition, preference of car over transit,

− 338 − KSCE Journal of Civil Engineering


Level of Service for On-street Parking

mode shift from car to public transit etc. may also be considered Vol. 51, No. 1, pp. 111-125.
for estimating the LOS, which have not been considered in the Moeinaddini, M., Asadi-Shekari, Z., Ismail, C. R., and Shah, M. Z.
present study. (2013). “A practical method for evaluating parking area level of
service.” Land Use Policy, Vol. 33, pp. 1-10, DOI: 10.1016/
j.landusepol.2012.11.014.
Acknowledgements Rome, P. and Mukherjee, S. (2015). “Car parking-a perennial problem
of kolkata.” International Journal of Science and Research (IJSR)
We take this opportunity to express my deep sense of gratitude ISSN: 2319-7064, Vol. 4, Issue 6, pp. 2583-2587.
to thanks Kolkata Municipal Corporation (KMA) for allowing us Saaty, T. L. and Kearns, K. P. (1985). “Analytical planning: The organization
to conduct various types of surveys at various study locations. of system.” (International series in modern applied mathematics and
We also want to thank M. Tech students, Mr. Sudeep Biswas and computer science, v. 7). Peegamon Press Ltd.
Mr. Saptarshi Sen for their help in data collection. Shaaban, K. and Pande, A. (2015). “Classification tree analysis of factors
affecting parking choices in Qatar.” Case Studies on Transport Policy.
Shiftan, Y. and Burd-Eden, R. (2000). “Modelling the response to
Notations parking policy.” Presented at TRB (Transportation Research Board)
2001 Annual Meeting, Washington, DC, DOI: 10.3141/1765-05.
AET = Average ease time Shoup Donald (2007). “Cruising for parking.” Available, http://shoup.
APF = Average parking fee bol.ucla.edu/CruisingForParkingAccess.pdf.
APSUR = Average parking space utilization rate Smith, A. (2013). “Parking utilization in neighbourhood shopping
AS = Average spacing centers on transit routes in san jose, california: are minimum parking
CBD = Central Business District requirements too high?.” Transportation Research Record: Journal
of the Transportation Research Board, (2359), pp. 27-35, DOI:
DC = Design Characteristics
10.3141/2359-04.
INR = Indian Rupee Smith, M. S. (1985). “Parking Standards.” Parking: The Magazine of
L= Availability of lightning the Parking Industry, Vol. 24, No. 4, (July-August, 1985).
LOS = Level Of Service Smith, M. S. (1996). Crime prevention through environmental design in
PTR = Parking turnover rate parking facilities, US Department of Justice, Office of Justice
PC = Parking Characteristics Programs, National Institute of Justice.
SC = Safety Characteristics Smith, M. S. and Thomas A. Butcher (2008). “How Far Should Parkers
V/C ratio = Volume to capacity ratio Have to Walk?.” Parking, Vol. 47, No. 4 .
Victoria Transport Policy Institute (2011). “Parking Evaluation—
VPFC = Number of vehicles per fee collector
Evaluating Parking Problems, Solutions, Costs, and Benefits”.
Report updated October 11, 2011.
References Yang, H., Hao, S., and Lei, W. (2010). “Research on the Improvement of
Urban Public Car Park Level of Service.” Transport Standardization
Cats, O., Zhang, C., and Nissan, A. (2016). “Survey methodology for May, pp. 87-90.
measuring parking occupancy: Impacts of an on-street parking Ye, X., Chen, J., Wang, T., Zheng, J., and Zhang, H. (2013). October.
pricing scheme in an urban center.” Transport Policy, Vol. 47, “Level of Service of Curbside Parking Facilities Based on Parking
pp. 55-63, DOI: 10.1016/j.tranpol.2015.12.008. Accessibility.” In Fourth International Conference on Transportation
Dong, H. Y. and Wang, Q. P. (2009). “The level of the car park service Engineering, DOI: 10.1061/9780784413159.318.
evaluation research based on fuzzy theory [J].” Technology & Yu, J. C. and Lincoln, A. R. (1973). “Parking Facility Layout: Level-of-
Economy in Areas of Communications, Vol. 6, pp. 14-17. Service Approach.” Journal of the Transportation Engineering
He, Y., Sun, X., Du, L., Jinmei, R., and Das, S. (2012). September. “Level of Division, Vol. 99, No. 2, pp. 297-306.
service for parking facilities.” In Intelligent Transportation Systems
(ITSC), 2012 15th International IEEE Conference on (pp. 1161-
1165). IEEE. DOI: 10.1109/ITSC.2012.6338914.
Appendix
Kay, B. and Smith, M. S. (2000). The level of service approach, France:
France Publications, Inc. Case study of Dalhousie area
Keese, C. L. (1964). “Highway Capacity the Level of Service Concept
(Doctoral dissertation, Texas A&M)”. Table 16. Collected Data for Each Variable from Field Survey for
Khanna, S. K. and Justo, C. E. G. (1991). Highway engineering, Nem Dalhousie
Chand & Bros. Variables 07:00 to 12:00 12:00 to 18:00 18:00 to 20:00
Kolkata Municipal Corporation (2000). Parking policy in Kolkata V/C ratio 0.90 1.33 0.84
Metropolitan Area (KMA).” Report, Available: http://www.cseindia. APSUR 0.48 0.94 0.64
org/userfiles/Parking%20Policy_KMDA.pdf. APF INR 68
Kononov, J. and Allery, B. (2003). “Level of service of safety: Conceptual AET 2 3 2
blueprint and analytical framework.” Transportation Research Record:
PTR 0.87 0.37 0.26
Journal of the Transportation Research Board, (1840), pp. 57-66,
VPFC 14 18 14
DOI: 10.3141/1840-07.
Mildner, G. C., Strathman, J. G., and Bianco, M. J. (1997). “Parking L Day paring Day parking 10.5
policies and commuting behaviour.” Transportation Quarterly, AS 1.10 0.16 1.70

Vol. 22, No. 1 / January 2018 − 339 −


Debasish Das and Mokaddes Ali Ahmed

Table 17. Dimensionless Value for Each Variable for Dalhousie Table 18. Average of Dimensionless Value for Each Parking
Variables 07:00 to 12:00 12:00 to 18:00 18:00 to 20:00 Selection Criteria for Dalhousie
V/C ratio 0.52 0.02 0.56 Parking Time
APSUR 0.67 0.08 0.48 characteristics 07:00 to 12:00 12:00 to 18:00 18:00 to 20:00
APF 0.47 PC 0.61 0.34 0.49
AET 0.80 0.75 0.80 SC 0.69 0.625 0.34
PTR 0.61 0.02 0.14 DC 0.63 0.14 0.89
VPFC 0.38 0.25 0.38
L 1 1 0.30 Table 19. Estimated Overall LOS for Dalhousie
AS 0.63 0.14 0.89 Time LOS value Designated LOS
07:00 to 12:00 0.63 LOS B
12:00 to 18:00 0.39 LOS C
18:00 to 20:00 0.51 LOS B

− 340 − KSCE Journal of Civil Engineering

You might also like