Incorporating lane blockages by trucks in intersection analysis and signal timing

Tronspn Rss. Vol. ISA. No. 6, pp. 45944, Printed in Great Britain.

1981

019I-2Mn/8l/06045SmrO2.00/0 01981 Pergamon Press Ltd.

INCORPORATING LANE BLOCKAGES BY TRUCKS IN INTERSECTION ANALYSIS AND SIGNAL TIMING PHILIP A. HABIB

Polytechnic Institute of New York, 333 Jay Street, Brooklyn, NY 11201,U.S.A. (Receioed 27 he

1980; in revised form 20 December

1980)

Abstract-This paper presents a modified method of intersection analysis that considers the operational effects of random lane blockages by the pickup and delivery vehicles that dominate the downtown arterial system. The paper provides a summary of the research that developed the presented tool, but concentrates on the findings related to mean arterial speed reduction caused by lane blockages. It then develops a method of determining an equivalent volume to represent the induced delays by the lane blockages. The paper subsequently presents a method of incorporating this equivalent volume effect into the standard critical lane method for signal timing. The author also makes suggestions on how the results could be used in other methods of intersection analysis.

The pickup and delivery (PUD) of freight at curbside on downtown arterials and streets has the potential of causing traffic interruption when the trucks double-park or otherwise block a through (moving) lane. Lane blockages by PUD vehicles are caused by three factors: (1) the vast majority of urban establishments do not have off-street loading facilities, (2) the PUD driver always seeks a stopping location very close (generally within 35 meters) to the ultimate destination and (3) necessary curbspace is not always available for parking without a lane blockage. The problem of truck lane blockages is concentrated in the downtown sections where competition for street space and curbspace is high. The street network in this area is predominantly one-way and traffic signals are located at nearly every intersection. Pedestrian activity is present in moderate to heavy amounts and local buses also use these downtown arterials. Due to the stresses of these competing users (and others), the downtown arterial does not seem capable of the “text book throughput” anticipated from that calculated in a conventional level of service/signal timing analysis. The purpose of this paper is to present a method of incorporating sparadic lane blockages, whether peak hour or off-peak, in intersection analysis on a dense one-way urban grid, usually that characteristic of the downtown of a larger “mature” city. The findings presented here are drawn from a larger project in curbside pickup and delivery of freight in urban areas. The analysis procedure presented herein is an end product of a portion of this larger project by Habib (1979) and in an effort to maintain brevity in this paper, several findings will be introduced and used without the usual necessary background development. BACKGROUND

The data on which this paper is based was collected in six cities: Boston, Dallas, Oklahoma City, San Francisco, St. Paul and Phoenix. The effect of actual truck-lane blockages was recorded on one second time-lapse photography. Ten such blockage case studies were used to calibrate NETSIM, a discrete traffic simulation model

(FHWA, 1980). One-way and two-way arterials were included in these case studies. After calibration, NETSIM was used to simulate the effects of different types (and combinations) of blockages that can occur on a one-way arterial. Blockages can occur at the upstream end of a block, at mid-block, at the downstream approach, or any simultaneous combination on either side of the street. The impact to traffic of each type of blockage is not unique. That is, the e#ect of a lane blockage at the intersection approach overshadows almost all other blockage types that may be simultaneously occurring. After review and analysis of all blockage combinations on a one-way block, Figure 1 defines six unique blockage configurations into which all other configurations may be represented (from the traffic impact viewpoint). The figure also shows the effect, in terms of speed reduction (on the blocked link), of each of these blockage configurations on varied arterial volume levels. Except for the approach-lane blockage (configuration l), the absolute value of the speed reduction peaks at about 1000 to 1100 vehicles per hour of green per lane. Above these values, the intersection increasingly controls speed and delay on the block (though never entirely except for a mid-block blockage). The following is a listing of the ranges of the simulation parameters done for the development of Fig. 1 results. Number of moving lanes Directions Volumes Blockage durations Intersection characteristics

Blocklength

2, 374 One-way v/c = 0.50, 0.70,0.80,0.85, 0.90 (5% trucks) 3,7, 12, 20 and 30 min 10% right and left turns, Green phase/cycle = 0.50, cycle length = 90 set 122m

It should be noted that the speed reduction values developed from the above simulations were for the link on which the blockage occurred. This would be some459

460

P. A. HABIB

*

* *

1

*

B

I

5

12.c

3 B ;

A

6.C

2 i? Y)

4.c

500

6 4 5 3 \2 1000

, 1500

h+sPLI

Fig. I. Volume versus speed reduction for lane blockage configurations what of an underestimate from the system perspective as delays would be felt on upstream blocks under the highest of volume conditions. Delays on other blocks, though acquired in the simulation are not considered, because of the effects of random lane blockages on these other blocks is not also imposed on the block of interest. ESTIMATION OF

EFFECT OF RANDOM BLOCKAGES

Given a total PUD demand “D” to a block (both sides of the street), then the percentage of double parkers “d” can be estimated using eqn (1) below. This equation is drawn from research by Habib (1979) on goods vehicle parking patterns. d = 0.49 - 0.41~

(1)

where p is the percentage of curbspace on the block “available” for goods movement. This consists of truck loading zones, fire zones adjacent to hydrants, bus stops, and other no-parking/no-standing zones. Thus dD is the hourly double-parking demand on the block. These double parkers will be distributed along both blockfaces. The analysis process divides the urban block into six cells (three cells on each side of the street) in which the PUD vehicles could double park. Assuming that the distribution of double parkers is uniform over the entire block, then the demand for double parking in each cell is dD/6 (see Fig. 2). The assumption on uniformity is a simplification of the analytics involved in

generating a solution. From the practical viewpoint, the distribution of double parkers on a block does tend to be more uniform than the PUD demand itself. The probability of any configuration state would be the product of the probabilities that double parkers only occupy the cells that define that configuration. On a typical urban blockface, each c’ell would be approximately 40 m long and capable of processing up to 3 PUD vehicles at one time. The mean dwell time of a double parker is about I I.5 min (Habib, 1979) excluding maneuvers, which can vary from one to three minutes. The analysis process conservatively assumes that each of the three serving positions in a cell can process PUD vehicles at a mean rate of four per hour. Therefore, the probability that any cell is empty (no double parkers in that cell) can be estimated by the general equation:

po= 2: (t)’ +i (~)c(--&) (2) where PO, is the probability of no PUD vehicles in any cell; A is the hourly arrival rate for double parkers to any cell; p is the service rate per position in any cell; C is the number of positions in any cell. For the analysis process developed herein, A = J!!? PUD/HR 6 y = 4 PUD/HR C = 3 positions per cell and

Fig. 2. Double-parking cell structure for PUD operations

where dD is the total number of double parkers arriving to the block per hour.

461

Intersection analysis and signal timing

The probability of any blockage configuration would then be calculated as follows: Prob. of configuration = (I - F’$ x F’g-”

(4)

where: b is number of cells in which there must be at least one PUD vehicle and PO is the probability that there is no PUD vehicle in the cell. As an example, the probability of configuration 2 (CJ on a one-way street (noting that this configuration can occur on either side of the street with the same impact) would be: P( C,) = 2 x (I - P&l5 The above technical presentation shows that the probability of an arterial block being in any of the various configuration states is dynamic and is a function of demand of PUD vehicles on that block. Therefore, the aggregate speed reduction AS due to the random transitioning from one blockage configuration state to another is: AS(V,)= iP(C,)xAS,(Vn) ,=I

(5)

where AS( V,) is the aggregate speed reduction for arterial traffic volume V,, P (Ci) is probability of configuration state i (which is also a function of demand), and A$( V,) is the speed reduction at traffic volume V, for configuration i (see Fig. I). Table I shows the results of the analysis for speed reductions at various arterial traffic volumes and number of blockages. It should be noted that volume is in vehicles per hour of green per lane and assumes 5% trucks, no local buses and moderate/heavy pedestrian crosswalk activity. DEVELOPlNGEQUlVALENTVOLUMES

For each NETSIM simulation done for the research, traffic was simulated under varied conditions before a lane blockage was introduced. As such, from the (calibrated) simulation data, a volume vs average speed “S” curve was developed to represent downtown arterial street conditions. Fig. 3 presents this curve. Table I. Speed reduction (KPH) for one-way arterials for various double-parking PUD vehicledemand on block-bothsides Number

of Double-Parker

Volume(Vo) VPHGPL

6 PUDS Per Hr.

Arrivals 12 PUDS Per Hr.

Per

Hour

24 PUDS Per Hr.

800

*

3.6

4.5

4.9

905

l

4.1

5.1

5.5

1000

l

4.7

5.5

5.9

1100

*

4.6

5.5

6.1

1200

l

4.5

5.5

6.2

1300

l

4.3

5.5

6.4

1400

*

4.2

5.5

6.6

Fig. 3. Volume-speed relationship

Figure 3 also shows the basics of the technique used to estimate the equivalent volume due to an average speed reduction of AS. If actual volume V, is known and if AS is determined from Table I (for a specified V, and number of blockages), then the equivalent volume V,, is estimated to be the volume whose operations results in an average speed of S-AS. This equivalent volume is read directly from Fig. 3. Table 2 is a summary of the results of findings relating actual and equivalent volumes for varied numbers of lane blockages on a one-way downtown arterial street. USEOFFINDINGSINANALYSIS

The goal of incorporating the effect of lane blockages in the intersection analysis/signal timing process is addressed by using the research results described herein to adjust upwards the intersection approach volume to account for the induced resistance. From the practical viewpoint, lane blockages by trucks alter intersection capacity by causing a suboptimal headway distribution, which in turn is caused by interrupted flow on the block plus an inefficient lane distribution of vehicles at the intersection approach. In keeping with the traffic engineering tradition of using adjustment factors to handle field conditions on an affected approach, this research defines a usable adjustment factor to account for random lane blockages. This is called the Blockage Factor and is defined as the ratio of Ve, to V, as presented in Table 2. Table 3 shows these values. In critical lane analysis, the lane utilization factor CJ takes into account the uneven distribution of through traffic over the various approach lanes. As the Blockage factor B developed herein attempts to aggregate all operational effects of random lane blockages (including the resultant uneven distribution of traffic by lane on the approach), to avoid double counting, B would replace U when conducting a critical lane analysis at the affected intersection. In other types of intersection capacity analysis, the Blockage Factor would be applied directly as another adjustment factor to the affected approach. As such, this paper presents new values to replace lane utilization for intersection approaches to reflect the adverse impact of these lane blockages. The method for

P. A. HABIB

462

Table 2. Equivalentvolume(V,,) of lane blockages Number (VO) (VPHGPL)

Volume

l

of Double-Parker

3 PUDS Per Hr.

Arrivals

24pIJDS Per Hr.

12 PUDS Per Hr.

6 PUDS Per Hr.

Per Hour

800

1016

1224

1272

1288

900

1080

1260

1305

1323

1000

1160

1310

1340

1360

1100

1232

1342

1375

1386

1200

1332

1380

1416

1440

1300

1430

IT*

FF

FF

1400

FF

FF

FF

FF

Based

on Forced

Flow

at approximately

1450

VPHGPL

Table 3. Recommended blockage factors (B) for downtown arterial streets Volume VPHGPL

Double 3

Parkers 6

Per Hour 12

24

800

1.05/1.10*

1.27

1.53

1.59

1.61

900

1.05/1.10

1.20

1.40

1.95

1.47

1000

1.05/1.10

1.16

1.31

1.34

1.36

1100

1.05/1.10

1.12

1.22

1.25

1.27

1200

1.05

1.08

1.15

1.18

1.20

1300

1.03

1.06

FF**

FF

FF

1400

1.01 ***

FF

FF

FF

FF

*

2 lane/3

** Forced l

l

**

lane

approach

Flow

** Suggested

values

finding and using these recommended values is presented

in the following example. Example

Problem: Determine the appropriate split for a signal at the following intersection: (also see Fig. 4): Major Street

Traffic Activity -400 vehicles per hour -I thru lane (12’) plus parking -moderate pedestrian activity along Minor St. -5% trucks -30% left turns -Peak hour factor = 0.91.

Goods Movement Activity on Approach Block: -18 trips in hour, 40% of all curb-space available for PUD Traffic Activity -1300 vehicles per hour -3 thru lanes (3 ~~12’)plus parking -heavy pedestrian activity along major street -5% trucks -10% right turns -Peak hour factor = 0.91. Minor Street

Goods Movement Activity on Approach Block: -None

Fig. 4. Example intersection

463

Intersection analysis and signal timing CONVENTIONALCRITICAL LANE ANALYSIS

Major Street Critical Lane Passenger Car Equivalents (4,) = [Vol. x (l-%TN) x F, x U]+ [Vol. x %TN x F, x Frl Lanes x PHF where Vol. is 1300 VPH; %TN, the percentage of right turns = 0.10; F, is a truck adjustment factor (0.95 + 2 X 0.05) = 1.05; U is the lane utilization factor = 1.10 for a 3 lane approach; F, is the turn pedestrian impedance factor = 1.50; Lanes is number of approach lanes = 3; PHF is the peak hour factor = 0.91. Solving the above equation, critical lane volume (4,) = 570 PCEPHPL.t Minor Street Passenger Car Equivalents per lane (&) = 496 PCEPHPL, where Vol. = 4OOvph; %TN = 0.30; Ff = 1.05; U = 1.0; F, = 1.25; Lanes = 1; PHF = 0.91. The sum of the critical lane volumes = 570 t 496 = 1,066 PCEPHPL which is level of service B according to the interim standards (Circular 212,198O). Assuming that 10% of the cycle time is lost to amber phases, the resultant phasing plan would be as follows: 4, = F

= 0.48

m,=!Z?&!!=O.42 Amber =

0.10 1.00

It is clear to the reader that the process as it is now

defined does not treat the reality of downtown goods movement and resultant lane blockages, which are characteristic of many downtowns, especially in the densest cities such as New York, San Francisco, Boston, Chicago and others. It is not accidental or a one time occurrence, but occurs every week-day on many blocks throughout the central areas. The analysis process as now defined is inadequate. RECOMMENDEDCRITICAL LANE ANALYSIS

The principal problem with the Blockage Factors presented in Table 3 is that the look-up process is based on VPHGPLS on the approach. This is a result of the analysis process used in their development. The solution to this inherent “problem” is brought out in the recommended analysis process outlined below: Step 1 Conduct the conventional critical lane analysis, without using the lane utilization factor U on any approach. Step 2 Calculate the resultant signal split. Step 3 Calculate the resultant VPHGPL for each approach. Step 4 Calculate the expected number of truck double parkers in the hour. Step 5 Find the Blockage Factor from Table 3. Step 6 Repeat the critical lane analysis using the respective Blockage Factor for all movements on all approaches. tpassenger car equivalent per hour per lane. SVehicle per hour of green per lane. ‘IX(A) I5:bc

Step 1

The conventional analysis, is repeated using U = 1.00. Therefore, Major Street Critical Lane Volume ($1) = 525 “PCEPHPL”. As there was no minor street lane utilization value, then Minor Street Equivalent Volume (&) = 496 “PCEPHPL” Step 2

Theoretical split: 4, = 0.46 42 = 0.44 0.10 Amber = 1.00 Step 3 = 942 VPHGPL

Major Street VPHGPL = &

As there is no goods movement activity on the minor street, subsequent analyses are not warranted. Step 4

Number of Double-Parkers on Major Street = Demand x(0.49-0.41p), where p is the portion of all curbspace available for curbside PUD activity (bus stops, loading zones, hydrant zones, driveways, other no parking/no standing zones). Number of Double-Parkers on Major Street = 18 (0.49 0.41 x 0.40) = 5.8 use 6.0 per hour. Step 5

From Table 3, using 942 VPHGPL and 6 blockages per hour, Blockage Factor = 1.36 Step 6 Major Street Critical Lane P.C.E. (4,) =

[vol x (l-%TN) x Ft x B] t [Vol x%TN x F, x F, x El Lanes x PHF where all variables are as defined previously and B is the Blockage Factor for that approach (in this case 1.36), Solving Major Street 4, = 714 PCEPHPL and Minor Street dZ = 496 PCEPHPL (unchanged) Sum of Critical lane volume is 1,210 PCEPHPL, which would be level of Service C. Recommended Final Phasing l#J= 0.53 f#Jz= 0.37 Amber

=

0.10 1.00

464

P. A. H~ere

Summary Downtown arterial street operations are a mixture of “activities” that standard intersection analysis tools cannot totally address. The author recognizes that the material presented herein is a first sfep to adopt improved tools to explain “urban congestion”. The use of the material presented primarily relates to volume ranges in the theoretical V/C range of 0.6-0.9, but congestion still results in various sections of the downtown. The thrust of the effort is to quantitatively show what additional major street green in needed over and above those determined from conventional analyses. The example presented shows that incorporating Blockage Factors in intersection analysis is a two step process. This two-step process is necessitated due to the realiza-

tion that the factors must be sensitive to traffic volume, which the research has demonstrated they are. Every downtown arterial street attracts PUD trips, and, depending on varied contributing factors such as curbspace availability, enforcement and driver attitudes, lane blockages become normal routine occurrences. The material presented in this paper recognizes this fact and offers a means of addressing it in intersection analysis. REFERENCES

Federal Highway Administration (1980) Trafic Network Analysis-A Users Guide. Washington, DC Habib P. A. (1979)Curbside Pickup and Delivery Operations and Arterial Trajic Impacts. Federal Highway Administration, Washington, D.C. Transportation Research Board (1980) Interim Mate&/s on Highway Capacity-Circular No. 212. Washington, D.C.