Professional Documents
Culture Documents
Chapter 38 - Network Analysis - 700
Chapter 38 - Network Analysis - 700
Chapter 38 - Network Analysis - 700
Suggested citation: National Academies of Sciences, Engineering, and Medicine. 2022. Highway
Capacity Manual 7th Edition: A Guide for Multimodal Mobility Analysis. Washington, DC: The
National Academies Press. https://doi.org/10.17226/26432.
The National Academy of Sciences was established in 1863 by an Act of Congress,
signed by President Lincoln, as a private, nongovernmental institution to advise the
nation on issues related to science and technology. Members are elected by their peers
for outstanding contributions to research. Dr. Marcia McNutt is president.
The National Academy of Engineering was established in 1964 under the charter of the
National Academy of Sciences to bring the practices of engineering to advising the
nation. Members are elected by their peers for extraordinary contributions to
engineering. Dr. John L. Anderson is president.
The National Academy of Medicine (formerly the Institute of Medicine) was established
in 1970 under the charter of the National Academy of Sciences to advise the nation on
medical and health issues. Members are elected by their peers for distinguished
contributions to medicine and health. Dr. Victor J. Dzau is president.
Learn more about the National Academies of Sciences, Engineering, and Medicine at
www.nationalacademies.org.
The Transportation Research Board is one of seven major programs of the National
Academies of Sciences, Engineering, and Medicine. The mission of the Transportation
Research Board is to provide leadership in transportation improvements and innovation
through trusted, timely, impartial, and evidence-based information exchange,
research, and advice regarding all modes of transportation. The Board’s varied
activities annually engage about 8,000 engineers, scientists, and other transportation
researchers and practitioners from the public and private sectors and academia, all of
whom contribute their expertise in the public interest. The program is supported by
state departments of transportation, federal agencies including the component
administrations of the U.S. Department of Transportation, and other organizations and
individuals interested in the development of transportation.
CHAPTER 38
NETWORK ANALYSIS
CONTENTS
1. INTRODUCTION.................................................................................................. 38-1
Overview ............................................................................................................. 38-1
Chapter Organization ........................................................................................ 38-1
Related HCM Content ........................................................................................ 38-1
5. REFERENCE.......................................................................................................... 38-80
LIST OF EXHIBITS
Exhibit 38-50 Example Problem 2A: Freeway Facility Geometric Features ...... 38-42
Exhibit 38-51 Example Problem 2A: Calculation of NB Right Turn
Capacity for a Single Cycle: Analysis Period 2 .............................................. 38-44
Exhibit 38-52 Example Problem 2A: NBR Capacity by Analysis Period ........... 38-44
Exhibit 38-53 Example Problem 2A: Calculation of the On-Ramp
Demand vR Based on the Intersection Operation .......................................... 38-45
Exhibit 38-54 Example Problem 2A: Freeway Facility Demand Inputs ............ 38-45
Exhibit 38-55 Example Problem 2A: Freeway Facility LOS ................................ 38-45
Exhibit 38-56 Example Problem 2A: Spillback Check: I-10 EB On-Ramp ......... 38-46
Exhibit 38-57 Example Problem 2A: Freeway Segment 5 Merge Capacity
and Queue Lengths ........................................................................................... 38-47
Exhibit 38-58 Example Problem 2A: Freeway Performance During
Analysis Period 4 with and without the Queue Storage Constraint .......... 38-47
Exhibit 38-59 Example Problem 2A: Estimated Queue Lengths and
Merge Capacities During Analysis Period 2 .................................................. 38-48
Exhibit 38-60 Example Problem 2A: Discharge Flow Rates into the On-
Ramp for Each Phase Throughout the Cycle During Analysis
Period 2 ............................................................................................................... 38-51
Exhibit 38-61 Example Problem 2A: Estimated Queue Lengths and
Merge Capacities During Analysis Period 3 .................................................. 38-52
Exhibit 38-62 Example Problem 2A: Discharge Flow Rates into the On-
Ramp for Each Phase Throughout the Cycle During Analysis
Period 3 ............................................................................................................... 38-54
Exhibit 38-63 Example Problem 2A: Calculation of the Spillback Capacity
Reduction Factor for the SBL Movement for Analysis Period 3 ................. 38-55
Exhibit 38-64 Example Problem 2A: Estimated Queue Lengths and
Merge Capacities During Analysis Period 4 .................................................. 38-56
Exhibit 38-65 Example Problem 2A: Calculation of the Spillback Capacity
Reduction Factor for the SBL Movement for Analysis Period 4 ................. 38-56
Exhibit 38-66 Example Problem 2A: Performance Measure Comparison
with and without Consideration of Spillback Effects ................................... 38-57
Exhibit 38-67 Example Problem 2B: TWSC Intersection Geometry: I-10 EB
Ramps .................................................................................................................. 38-58
Exhibit 38-68 Example Problem 2B: Calculation of the On-Ramp Demand
vR Based on the Intersection Operation .......................................................... 38-59
Exhibit 38-69 Example Problem 2B: Queue Accumulation Plot
Calculations for the On-Ramp ......................................................................... 38-60
Exhibit 38-70 Example Problem 2B: Queue Accumulation Polygon for the
On-Ramp............................................................................................................. 38-60
Exhibit 38-A26 Illustration of Mainline Flow Rate Split into Blocked and
Unblocked Lanes .............................................................................................38-111
Exhibit 38-A27 Procedure for Evaluating the Impact of Queue Spillback
on Upstream Nodes and Determining the Queue Length within
Upstream Segments .........................................................................................38-118
Exhibit 38-A28 Potential Effects of an Off-Ramp Queue on Node i ................38-119
Exhibit 38-A29 Distribution of pi as Function of Distance from the
Diverge Point, for a 3-Lane Segment ............................................................38-120
Exhibit 38-A30 Illustration of Lane-Change Maneuvers within the Queue
Influence Area in a 4-Lane Segment under Regime 3 .................................38-121
Exhibit 38-A31 Illustration of Lane-Change Maneuvers within the Queue
Influence Area in a 4-Lane Segment under Regime 4 .................................38-121
Exhibit 38-A32 Effect of Queue Spillback on the Discharge Capacity of an
Upstream On-Ramp ........................................................................................38-123
Exhibit 38-A33 Illustration of Different Density Values within One
Diverge Segment ..............................................................................................38-124
Exhibit 38-B1 Procedure for Detecting Spillback Occurrence at an On-
Ramp .................................................................................................................38-131
Exhibit 38-B2 Schematic of Movements Turning to an On-Ramp from a
TWSC Intersection ...........................................................................................38-133
Exhibit 38-B3 Schematic of Movements Turning to an On-Ramp from an
AWSC Intersection ..........................................................................................38-135
Exhibit 38-B4 Schematic of Movements Turning to an On-Ramp from a
Roundabout ......................................................................................................38-135
Exhibit 38-B5 Signalized Intersection Methodology With Adjustments to
Address On-Ramp Queue Spillback .............................................................38-138
Exhibit 38-B6 Typical Signalized Intersection Ramp Terminal in a
Diamond Interchange .....................................................................................38-139
Exhibit 38-B7 Estimation of Freeway On-Ramp Merging Capacity ................38-140
Exhibit 38-B8 Sample Intersection for Calculation of a QAP for the On-
Ramp .................................................................................................................38-141
Exhibit 38-B9 On-Ramp Queue Accumulation Polygon During Queue
Spillback ............................................................................................................38-141
Exhibit 38-B10 Illustration of Cooperative Behavior in Unsignalized
Intersections with Queue Spillback ...............................................................38-143
Exhibit 38-B11 TWSC intersections Core Methodology with Adjustments
to Address On-Ramp Queue Spillback .........................................................38-144
Exhibit 38-B12 On-Ramp Queue Accumulation Polygon: TWSC
Intersection .......................................................................................................38-145
Exhibit 38-B13 AWSC Intersection Core Methodology with Adjustments
to Address On-Ramp Queue Spillback .........................................................38-148
1. INTRODUCTION
CHAPTER ORGANIZATION
Section 2 provides the performance measures used at the network level and An O-D pair represents the
includes example calculations of O-D travel time and network travel time. route between two specific
points in the analysis network.
Section 3 describes procedures to evaluate spillback impact on a freeway due Points are defined in Chapter
2, Applications.
to congestion on a ramp or urban street.
Section 4 describes procedures to evaluate spillback impact on an urban
street due to congestion on the freeway or on-ramp.
Section 5 provides case studies to illustrate the application of this chapter’s
methods.
A series of appendices provide detailed information about specific models
and analysis steps.
• Chapters 19, 20, 21, and 22, which provide analysis tools for signalized
intersections, two-way STOP-controlled intersections, all-way STOP-
controlled intersections, and roundabouts, respectively; and
• Chapter 23, which provides methods for evaluating ramp terminals and
alternative intersections.
2. CONCEPTS
OVERVIEW
This section discusses concepts related to spillback on the freeway, spillback
on the urban street, lane-by-lane analysis, and performance measurement for
networks and O-Ds. Concepts related to freeway analysis and urban street
analysis are described in their respective chapters elsewhere in the HCM.
Off-Ramp Elements
A freeway off-ramp typically consists of three components, as illustrated in
Exhibit 38-1.
• Deceleration lane(s), measured from the beginning of the taper of the
auxiliary lane to the gore.
• Ramp roadway, connecting the deceleration lane and the downstream
ramp terminal and measured from the gore to the taper of the ramp
terminal.
• Ramp terminal, connecting the ramp roadway to the urban street facility
and measured from the point where additional lanes are added to the
intersection approaches to the stop bar of the approach. This component’s
length should be at least as long as the approach’s turn bay lengths. The
ramp terminal can be uncontrolled, STOP- or YIELD-controlled, or
signalized.
When the ramp connects two freeway facilities, the downstream ramp
terminal is replaced by the merge section of the on-ramp, with no storage length.
Exhibit 38-1
Off-Ramp Components
Exhibit 38-2
Definition of Spillback
Regimes
(b) Regine 1: Queue within the deceleration lane (c) Regime 2: Queue along the shoulder
(d) Regime 3: Queue in the rightmost lane (e) Regime 4: Queue blockage of the adjacent lane
Regime 0
Under this regime, shown in Exhibit 38-2(a), there are no queues in the ramp
roadway or the queue, if it exists, is contained within the ramp roadway
boundaries. There are no operational effects in the ramp influence area.
Regime 1
Under this regime, shown in Exhibit 38-2(b), the queue ends within the
deceleration lane and does not spill back into the mainline freeway. Deceleration
lanes typically serve as a transition zone between speeds on the mainline
(typically 55–75 mi/h) and advisory speeds posted along the off-ramp roadway
(typically 20–50 mi/h). When queues begin to form on the deceleration lane, the
Regime 2
Under this regime, shown in Exhibit 38-2(c), the queue of vehicles extends
upstream beyond the deceleration lane, but sufficient lateral clearance on the
right-hand shoulder allows for additional queue storage. In this case, the
deceleration lane does not serve as a transition zone and drivers decelerate and
join the back of the queue more abruptly, resulting in turbulence and reduced
speeds in the rightmost lane. If no lateral clearance exists immediately upstream
of the deceleration lane, Regime 2 conditions are not possible. In some cases, this
regime does not occur even if storage is available; this occurrence is site-specific
and depends on local driver behavior.
Regime 3
Under this regime, shown in Exhibit 38-2(d), the queue extends to the
rightmost lane of the freeway mainline. This regime may occur either when no
shoulder is available for additional queue storage, or when drivers choose to
queue in the rightmost lane once the deceleration lane is entirely occupied. Non-
exiting vehicles on the rightmost lane are delayed or change lanes, which causes
increased turbulence and reduced speeds in the two rightmost lanes.
Regime 4
Under this regime, shown in Exhibit 38-2(e), the queue blocks the rightmost
lane, and drivers occasionally or often use the adjacent freeway mainline lane
next to the rightmost freeway mainline lane to force their way into the queue,
thus blocking an additional lane. During this regime, mainline speed and
capacity are significantly reduced.
Exhibit 38-4
Queue Influence Area with
Increased Turbulence
The length of the QIA is estimated as function of the segment free-flow speed
Additional discussion on the
determination of the Queue (FFS), as shown in Exhibit 38-5. During undersaturated operations, drivers have
Influence Area (QIA) is adequate warnings about the presence of a ramp through signage and navigation
presented in Appendix A.
aids and can position themselves according to their destination. However, when
queue spillback occurs ,drivers can only detect a downstream queue visually and
therefore have less time to react, resulting in more aggressive lane changes and
additional turbulence.
Exhibit 38-5 Segment Free-Flow Speed (mi/h) Queue Influence Area (ft)
Length of Queue Influence 50 810
Area as a Function of the 55 900
Segment Free-Flow Speed 60 980
(FFS) 65 1,060
70 1,140
75 1,220
Exhibit 38-6
Queue Spillback from an On-
Ramp into Urban Street
Intersections
The on-ramp queue length also depends on the upstream demands. In the
example shown in Exhibit 38-6(a), three movements contribute to this demand:
northbound (NB) right, southbound (SB) left, and eastbound (EB) through. If the
NB right movement is very heavy or has the right-of-way for a significant
amount of time, the SB left movement may not have much opportunity to
contribute to the demand and may spill back upstream, affecting the adjacent SB
through movement as well as the upstream intersection. Thus, in the case of
signalized intersections, the relative contribution of demands to the queue length
will depend on the relative demands of these movements and the respective
signal timings and right-of-way allocation. The discharge rate of these upstream
intersection movements will depend on the on-ramp’s storage availability during
the respective signal phase. The analysis estimates the additional lost time due to
the presence of the downstream queue and adjusts the effective green of the
affected movements.
In the roundabout example shown in Exhibit 38-6(b), the same three
movements contribute to the on-ramp demand. However, in this case, the
movements have priority in the following order: (a) SB left, (b) EB through, and
(c) NB right. A high-priority movement with a heavy demand may constrain the
entry capacity of lower-priority movements, resulting in total throughput that is
lower than the sum of the three contributing movement demands.
LANE-BY-LANE ANALYSIS
Spillback affects each lane of a facility differently. For example, when
spillback occurs at a freeway off-ramp, the rightmost lanes of the freeway may be
blocked, while the leftmost lanes operate in free-flow conditions. Therefore, the
methodology estimates operating conditions by lane as well as by segment. The
lane-by-lane performance metrics are also used to obtain O-D–based travel times.
The lane-by-lane analysis provides lane flow ratios (LFRs) representing the The demand flow rates by lane
are estimated as a percentage
percentage of the entering demand by lane. The LFR is a function of the segment- of the segment demand.
wide volume-to-capacity (v/c) ratio and values are provided for each segment
type (basic, merge, diverge, and weaving). In addition, FFS, speed, and capacity
are estimated by lane. When the facility becomes oversaturated, speeds are
estimated using the method of Chapter 10, Freeway Facilities, which is based on
interactions between successive segments.
3. METHODOLOGY
Performance Measures
The methodology of this chapter generates the following performance
measures:
• Freeway facilities:
o Flow, free-flow speed (FFS), operating speed, and capacity for
individual lanes
o Expected travel speed along each segment
• Urban street facilities:
o Travel time along each segment
o Expected travel speed along each segment
• Networks:
o Total and free-flow travel times
o Travel time index (TTI)
o Average speed
Limitations
The methodology has the following limitations:
1. Multiple overlapping breakdowns or bottlenecks cannot be fully
evaluated by this methodology. Consult Chapter 6, HCM and Alternative
Analysis Tools, for a discussion of simulation and other models.
2. Demand is an input into the process, and the methodology does not
address any changes in demand that are due to traffic operation conditions.
3. Managed lanes can be analyzed as part of the freeway system. However,
the interaction of managed lane operations with spillback conditions are
not addressed.
4. The methodology does not explicitly consider alternative intersection and
interchange designs, such as diverging diamond interchanges (DDIs) and
single-point urban interchanges (SPUIs). However, it can be extended to
consider these designs, assuming turning movements, demands, and
queues can be accurately estimated for the movements of interest.
5. The methodology does not consider two-lane roundabouts and their
interaction with freeway on-ramps.
6. The HCM’s reliability methods cannot be applied to network analysis
because the process for developing reliability scenarios is different for
freeways and arterials.
Required Input Potential Data Source(s) Suggested Default Value Exhibit 38-7
Trajectory Parameters by O-D Pair Required Input Data, Potential
Data Sources, and Default
Origin and destination points Set by analyst Must be provided
Values for the Network
Route between origin and destination Set by analyst Must be provided
Analysis Methodology
Freeway Facilities
Input data for facility and segment As shown in Chapters 10, As shown in Chapters 10, 12,
methods 12, 13, and 14 13, and 14
Ramp access density
Road geometry Must be provided
(number of ramps within 1 mi)
Grade (%) Road geometry Must be provided
Urban Street Facilities
Input data for facility and segment As shown in Chapters 16 As shown in Chapters 16 and
methods and 18–23 18–23
Urban street segments: corresponding Set by analyst, according to
Must be provided
movement at downstream intersection the selected route
Off-Ramp Queue Spillback
Off-ramp queue spillback: Road geometry, Function of diverge geometry
expected number of queued lanes field observations and driver aggressiveness
Length of available shoulder (ft) Road geometry Must be provided
Off-ramp detailed geometry Road geometry Must be provided
On-Ramp Queue Spillback
On-ramp metering rate (veh/h)
Field data Must be provided
(if applicable)
On-ramp detailed geometry Road geometry Must be provided
Roundabouts: exit capacity (pc/h) Field data, past counts 1,300 pc/h
COMPUTATIONAL STEPS
This section describes the methodology’s computational steps. Exhibit 38-9
illustrates the process used to evaluate network operations.
Exhibit 38-9
Network Analysis Methodology
Flowchart
Exhibit 38-11
Potential Bottlenecks
Constraining the Ramp
Terminal Demand
𝑣𝑅 𝑐𝑅
𝑣𝑖, 𝑎𝑑𝑗 = 𝑣𝑖 × × min ( , 1) Equation 38-2
∑𝑖 𝑣𝑖 𝑣𝑅,𝑎𝑑𝑗
where
vi,adj = adjusted demand for movement 𝑖 at the downstream intersection
(pc/h);
vi = demand for movement 𝑖 at the downstream intersection (pc/h);
vR = off-ramp demand (pc/h);
vR,adj = adjusted off-ramp demand, from Equation 38-1) (pc/h); and
cR = capacity of ramp roadway, from Exhibit 14-12 (pc/h).
Exhibit 38-12
Potential Bottlenecks
Constraining the On-Ramp
Demand
However, if capacity is exceeded at any of those locations, the flow that will
reach the freeway merge will be lower than the on-ramp demand vR and
adjustments should be made to the respective volumes.
If any of the ramp terminal movements that discharge into the on-ramp
operate over capacity, the total throughput to the on-ramp is given by
𝑁
1
𝑣𝑅,𝑎𝑑𝑗 = × ∑ min(𝑣𝑖 , 𝑐𝑖 ) Equation 38-3
𝑓𝐻𝑉
1
where
To keep the units of vR,adj = adjusted on-ramp demand (pc/h);
intersection throughput
(veh/h) and freeway on-ramp
demand (pc/h) consistent,
vHV = heavy vehicle adjustment factor (decimal), from Equation 12-10;
Equation 38-3 applies the
heavy vehicle adjustment vi = demand for movement i at the intersection (veh/h);
factor given by Equation 12-10.
ci = capacity of movement i at the intersection (veh/h); and
N = number of intersection movements that discharge into the on-ramp.
If the total on-ramp demand vR is greater than the ramp roadway capacity cR,
the adjusted on-ramp demand is:
Equation 38-4 𝑣𝑅,𝑎𝑑𝑗 = min(𝑣𝑅 , 𝑐𝑅 )
where
vR,adj = adjusted on-ramp demand (pc/h);
vR = on-ramp demand (pc/h); and
cR = ramp roadway capacity (pc/h), from Exhibit 14-12.
Exhibit 38-13
Spillback Check Procedure for
Off-Ramps
Exhibit 38-14
Spillback Check Procedure for
On-Ramps
Step 5A: Compute Operating Speeds for Individual Lanes Along the
Freeway Facility
The operational performance of a freeway facility is determined from the
density and speed of each segment along the facility. The average travel time for
a given segment can be derived from its average speed.
For a network analysis, the segment speed is function of:
• Estimated speeds for individual lanes, and
• Probability that a lane will be selected for the subject O-D pair.
To estimate the speeds and capacities for individual lanes, a set of models
have been developed for each type of freeway segment considering the total
number of mainline freeway lanes. These models are valid only for
undersaturated conditions, and they predict the lane flow ratio (LFR) for each
lane. These models are of the form:
𝐿𝐹𝑅𝑖 = 𝑎 × ln(𝑣/𝑐) + 𝑏 Equation 38-5
𝑁−1
For all segment types, the share of flow is estimated on the mainline
upstream of the segment. The oversaturated portion of the speed–flow curve
(when density is greater than the density at capacity) cannot be addressed by the
speed–flow models, as this is a limitation of the existing methods. The lane-by-
lane flows for oversaturated conditions are estimated using the procedures of
Chapter 25, adjusted to determine the incoming and outgoing flow on a lane-by-
lane basis. However, if off-ramp queue spillback occurs in the freeway facility,
then the methodology in Appendix A is used to determine the lane-by-lane flow
distribution.
The probability that a given lane is selected when traveling between a given
O-D pair depends on the segment’s location. For segments where a motorist
driving between the O-D pair would enter (merge segment) or leave the freeway
facility (diverge segment), the probability of lane selection is shown in Exhibit 38-
15 (assuming right-side ramps).
For other segments within the freeway facility, the probability pi,j of choosing
a given lane i on segment j is equal to the lane’s LFR, defined as the percentage of
the total flow assigned to lane i:
Equation 38-8 𝑝𝑖,𝑗 = 𝐿𝐹𝑅𝑖,𝑗
This concept is illustrated in Exhibit 38-16 for a 3-lane freeway facility with
nine segments. The exhibit shows the lane choice probabilities for the O-D pair
where the traveler enters the freeway facility on segment 2 (merge) and leaves
the freeway on segment 8 (diverge). For segments 2 and 8, the choice
probabilities for lanes 1, 2 and 3 are 0.90, 0.05 and 0,05 respectively, from Exhibit
38-15. For segments 3 through 8, the lane choice probabilities are equal to the
LFR (Equation 38-3), calculated for each lane of each segment.
Exhibit 38-16
Illustration of Lane Choice
Probabilities Along a Freeway
Facility
Each segment’s speed is then computed as the sum of products of speeds for
each lane and the corresponding probability of lane choice:
𝑁
where
Se,j = expected speed for segment j (mi/h),
N = total number of lanes in the segment,
pi,j = probability that lane i is selected (decimal) on segment j, and
Si = speed of lane i (mi/h), from Equation 38-C12.
A special case occurs when a weaving segment is defined as an entry or exit
segment. The methodology for estimating lane-by-lane speeds cannot evaluate
auxiliary lanes in weaving segments; therefore, the expected speed equation
should be replaced by the average speed of weaving vehicles in the segment.
Additional details are provided in Appendix C: Lane-by-Lane Analysis for
Freeway Facilities.
The speed–flow relationship for ramps is linear and speed decreases with
higher ramp flows, as shown in Exhibit 38-17. The maximum allowed values of
vR are bounded by ramp capacity, consistent with guidance provided by Chapter
14, Freeway Merge and Diverge Segments (Exhibit 14-12).
Exhibit 38-17
Speed–Flow Curves for
Freeway Ramps
The travel time along freeway ramps is calculated by dividing the ramp
length by its respective speed. When an O-D pair includes an off-ramp, the
control delay for the corresponding movement at the at-grade intersection must
also be added to the off-ramp travel time. This calculation is consistent with the
urban street facility methods, where each segment’s travel time includes the
control delay of the corresponding movement at the downstream intersection.
For off-ramps:
𝐿𝑅
Equation 38-12 𝑇𝑇ramp = + 𝑑𝑖 + 5
1.47 × 𝑆ramp
For on-ramps:
𝐿𝑅
Equation 38-13 𝑇𝑇ramp = +5
1.47 × 𝑆ramp
The constant value of 5 in where
Equation 38-12 and Equation
38-13 accounts for delay due TTramp = ramp travel time (s),
to vehicle deceleration and
acceleration along off-ramps Sramp = ramp speed (mi/h),
and on-ramps, respectively.
LR = ramp length (ft), and
di = control delay at the downstream ramp terminal for the O-D pair’s
corresponding movement (s) (applicable for off-ramps only).
In the case of queue spillback, Appendix A describes the procedure for
estimating the ramp speed, which is similar to the Oversaturated Segment
Evaluation method described in Chapter 25.
Step 8: Compute Travel Times for the Network and Each O-D Pair
This step computes the total travel time TTO-D for the network as the sum of
the travel times over all segments along the route. For multiperiod analysis, it is
important to select the travel time for the correct analysis period at each segment,
as a long O-D route may encompass several analysis periods. Exhibit 38-18
presents a sample calculation for a facility with two analysis periods (15 min
each). The first segment in the O-D route is traversed during analysis period 1,
and the cumulative travel time column is updated with the respective value.
Subsequent segments follow the same procedure until the cumulative travel time
exceeds the length of the first analysis period (900 s). For the next segment in the
network, travel times from analysis period 2 are added to the cumulative travel
time column. This procedure is then repeated until the final segment is reached.
The total travel time is obtained as the last value in the cumulative travel time
column.
Freeway Facilities
At free-flow, the speed of freeway segments is computed as being equal to
their free-flow speed. When a lane-by-lane analysis is applied, the methodology
computes the free-flow speed for each lane (Equation 38-C9).
Next, the probabilities of lane choice on each segment are calculated for each
segment. If the subject segment is a entry or exit segment (segments where the
driver on a particular O-D route enters or leaves the freeway facility, as
illustrated in Exhibit 38-16), the lane choice probabilities are obtained from
Exhibit 38-15. For other segments, the lane choice probability is equal to the LFR
(Equation 38-5). When calculating the LFR under free-flow conditions, a v/c value
of 0.1 is recommended to provide results consistent with field data. Due to the
logarithmic form of the LFR equation, using v/c = 0 is mathematically infeasible,
and very low v/c values yield unrealistic results.
travel speed that accounts for congestion is the control delay for the specific O-D
related movement at the downstream intersection, which is discussed next.
Signalized Intersections
The control delay for a given lane at a signalized intersection is provided by
Equation 19-18, repeated here as Equation 38-18:
Equation 38-18 𝑑 = 𝑑1 + 𝑑2 + 𝑑3
where
d = control delay (s/veh);
d1 = uniform delay (s/veh), from Equation 19-19;
d2 = incremental delay (s/veh), from Equation 19-26; and
d3 = initial queue delay (s/veh), from Equation 19-44.
Under free-flow conditions, the values of d2 and d3 are equal to zero.
Therefore, the free-flow control delay is equal to the value of uniform delay (d1)
computed for a demand-to-capacity ratio X of 0.
3,600 𝑣
3,600 𝑣𝑥 𝑣
2 ( )( 𝑥 )
𝑐 𝑐
− 1 + √(
𝑥 𝑚,𝑥 𝑚,𝑥
Equation 38-19 𝑑= + 900𝑇 − 1) + +5
𝑐𝑚,𝑥 𝑐𝑚,𝑥 𝑐𝑚,𝑥 450𝑇
[ ]
where
d = control delay (s/veh);
vx = flow rate for movement x (veh/h);
cm,x = capacity of movement x (veh/h), from Chapter 20; and
T = analysis time period (0.25 h for a 15-min period) (h).
Under free-flow conditions, the demand vx is set to zero, which allows
Equation 38-19 to be reduced to the following:
3,600
Equation 38-20 𝑑= +5
𝑐𝑚,𝑥
where all variables are as defined previously.
ℎ𝑑 𝑥
𝑑 = 𝑡𝑠 + 900𝑇 [𝑥 − 1 + √(𝑥 − 1)2 + ]+5 Equation 38-21
450𝑇
where
d = average control delay (s/veh);
x = vhd/3,600 = degree of utilization (unitless);
ts = service time (s), from Equation 21-29;
hd = departure headway (s), from Equation 21-28; and
T = length of analysis period (0.25 h for a 15-min period) (h).
Under free-flow conditions, the degree of utilization x is set to zero, which
allows Equation 38-21 to be reduced to the following:
𝑑 = 𝑡𝑠 + 5 Equation 38-22
Roundabouts
The control delay for roundabouts is computed by Equation 22-17, repeated
here as Equation 38-23:
3,600
3,600 √(𝑥 ( 𝑐 )𝑥
𝑑= + 900𝑇 [𝑥 − 1 + 2
− 1) + ] + 5 × min[𝑥, 1] Equation 38-23
𝑐 450𝑇
where
d = average control delay (s/veh);
x = volume-to-capacity ratio of the subject lane;
c = capacity of the subject lane (veh/h), from Equation 22-14; and
T = time period (h) (T = 0.25 h for a 15-min analysis).
Similar to TWSC intersections, setting the volume-to-capacity ratio x to 0
reduces Equation 38-23 to a simpler form:
3,600
𝑑= Equation 38-24
𝑐𝑚,𝑥
Freeway Ramps
Freeway ramp speeds at free-flow are equal to the ramp free-flow speed SFR
provided by the analyst and do not require additional adjustments.
4. EXAMPLE PROBLEMS
This section presents four example problems (Exhibit 38-20) illustrating the
evaluation of networks and addressing several cases of spillback onto freeways
and onto urban street facilities.
Exhibit 38-21
Example Problem 1: Network
Interchanges, Intersections,
and O-D Points
The analysis steps for evaluating this network are discussed below.
The average total travel time between each O-D pair can be obtained by
adding the average travel times on each segment and ramp roadway traversed,
plus the average delay experienced at all intersections along the route.
The O-D route from node D to node H will traverse two urban street
facilities, as shown in Exhibit 38-24:
• Archer Rd. westbound (WB), consisting of two urban street segments and
two signalized intersections (SW 40th Blvd. and I-75 NB on-ramp); and
• NW 39th Ave. eastbound (EB), consisting of one urban street segment and
two signalized intersections (I-75 NB off-ramp and NW 95th Blvd).
Exhibit 38-24
Example Problem 1:
Urban Street Facilities
(a) Urban Street Facility 1: Archer Rd. WB (b) Urban Street Facility 2: NW 39th Ave. EB
Notes: The movements whose control delay will be used in the analysis are indicated in parentheses.
WBR = westbound right turn, WBT = westbound through, NBR = northbound right turn, EBT = eastbound
through.
The O-D route also includes the freeway facility (I-75 NB), starting at
segment 8 and ending at segment 16, as shown in Exhibit 38-22. The on-ramp
and off-ramp at the boundary ends of the facility are also included in the travel
time evaluation.
Exhibit 38-25 lists the segments, ramps, and intersections traversed for the
route connecting O-D pair D-H.
The analysis’ temporal scope must also be defined. Given the short length of
the subject network, a single-period analysis will be performed initially. The
average travel time obtained will be compared to the 15-min analysis period
length; if the travel time is longer than 15 min, the study’s temporal scope will be
reevaluated.
Archer Rd. @ Demand (veh/h) 120 2,348 88 36 864 548 60 208 96 36 480 304
SW 40th Blvd. Phase split (s) 20 50 — 20 50 — 20 30 — 20 30 —
Note: L = left, T = through, R = right.
Additional input parameters for Urban Street Facilities 1 and 2 are as follows:
• Base saturation flow rate: 1,900 veh/h/ln
• Traffic composition: 0% heavy vehicles
• Cycle length: 120 s
• Grade: 0%
• Arrival type: 3
• Speed limit: 45 mi/h
• Yellow change interval: 4 s
• Red clearance interval: 0 s
• No pedestrians
Next, the off-ramp volumes are checked against the intersection turning
movement demands. The first check determines whether there are bottlenecks
along the freeway facility that may meter off-ramp demands. Exhibit 38-32
shows the estimated LOS for all 19 segments in the freeway facility. Since no
segment is oversaturated, the off-ramp demand is not metered, and no
adjustments are necessary.
The second check compares the off-ramp demands to the respective ramp
roadway capacity, as shown in Exhibit 38-33. Demand does not exceed capacity
for any of the ramps; therefore, no adjustments to the intersection volumes are
performed.
• Queue storage ratio: Any queues exceeding the available turn bay length at
the intersection must be checked against the available storage along the
ramp roadway. For single-lane off-ramps, any queues upstream of the
intersection will share the same storage and must be aggregated. If a
ramp has two or more lanes, the analyst must determine how ramp lanes
are channelized relative to intersection approaches, based on the off-ramp
geometry. As shown in Exhibit 38-33, only the off-ramp at segment 16
(NW 39th Ave.) has two lanes—the leftmost ramp lane L2 is connected to
the left-turn movement, while the rightmost ramp lane L1 is connected to
the right-turn movement. In this step, the only movement that must be
evaluated is the left turn at Williston Rd. The queue length upstream of
the intersection is compared to the available ramp length, with a resulting
queue storage ratio RQ = 439 / 900 = 0.49 < 1.0. Therefore, spillback is not
expected to occur along the off-ramps.
Step 5A: Obtain Speeds for Individual Lanes in the Freeway Facility
First, the flow distribution among freeway lanes must be determined for the
segments in the freeway facility. Using the estimated flow rates, lane speeds are
computed as shown in Exhibit 38-35. The highlighted rows (8 through 16)
represent the segments included in the O-D route and used to compute the
overall travel time. The rightmost lane is labeled Lane 1.
Segment Segment Lane Flow Ratio Lane Speed (mi/h) Exhibit 38-35
ID LOS Lane 1 Lane 2 Lane 3 Lane 1 Lane 2 Lane 3 Example Problem 1:
1 C 0.286 0.351 0.362 66.5 71.3 77.7 Flow Distribution and Speeds
2 C 0.338 0.319 0.343 56.7 73.1 77.5 for Freeway Segments
3 C 0.279 0.356 0.365 68.3 71.6 77.2
4 C 0.259 0.388 0.353 72.0 72.4 77.3
5 C 0.281 0.348 0.371 66.4 69.9 75.6
6 C 0.336 0.326 0.337 55.3 72.2 77.4
7 C 0.286 0.354 0.360 67.7 70.6 76.5
8 C 0.253 0.387 0.359 71.7 71.5 76.8
9 C 0.294 0.346 0.360 56.3 67.3 74.0
10 C 0.288 0.344 0.368 58.2 67.7 73.9
11 D 0.358 0.290 0.352 41.6 71.7 75.2
12 C 0.286 0.355 0.359 67.8 70.8 76.6
13 B 0.253 0.382 0.365 71.9 72.1 76.8
14 C 0.281 0.349 0.370 66.8 70.0 75.7
15 C 0.281 0.349 0.370 66.8 70.0 75.7
16 C 0.350 0.296 0.354 50.6 74.1 76.8
17 B 0.278 0.362 0.361 68.3 72.6 78.1
18 B 0.252 0.383 0.365 72.0 74.2 77.7
19 B 0.272 0.358 0.370 68.7 72.7 78.1
Note: Bold rows represent the segments included in the O-D route that are used to compute the total travel time.
The step by step calculations
to determine lane-by-lane
Next, the expected speed for each segment is computed as the sum of flows and speeds on segment
products of speeds for each lane and the corresponding probability of lane 16 (diverge) are presented in
Example 1 in Appendix C.
choice, as provided in Equation 38-9. The results are shown in Exhibit 38-36.
Even though the remaining segments’ travel times segments are not directly
used in calculating the O-D travel time, the entire facility must be analyzed, as
any existing bottleneck would affect the performance of other segments along the
facility. In this example, no segment operates at LOS F, and no queues develop at
the ramps connecting to urban streets.
Exhibit 38-39 Segment ID Expected Speed (mi/h) Segment Length (ft) Travel Time (s)
Example Problem 1: Freeway 8 68.4 1,500 15.0
Segment Travel Times 9 66.5 6,300 64.6
10 67.2 5,385 54.6
11 64.4 1,500 15.9
12 72.0 2,014 19.1
13 73.8 1,500 13.9
14 71.2 6,494 62.2
15 71.2 2,480 23.7
16 53.1 1,500 19.3
Step 8: Compute Travel Times for the Network and Each O-D
All segments within the subject O-D route (D-H) are sorted according to their
travel sequence and their respective travel times are listed as shown in Exhibit
38-41. The cumulative travel time for the O-D route must also be computed to
evaluate whether the network analysis is being contained correctly within the
temporal scope defined in Step 1. For this example, a single 15-min analysis
period was considered, for a total time of 900 s. Because the cumulative travel
time does not exceed this boundary value, all travel times obtained from analysis
period 1 are valid for the analysis.
Exhibit 38-42
Example Problem 2: Network
Interchanges, Intersections,
and O-D Points
The freeway facility (I-10 EB) will be modeled according to the freeway
facilities core methodology (Chapter 10), while the ramp terminal will be
modeled according to its respective intersection methodology (Chapter 19, 20, or
21). First, a check will be performed to confirm the occurrence of queue spillback.
Next, the respective spillback analysis method will be applied to evaluate the
impacts of queue spillback on the capacity of each movement at the intersection.
These reduced capacities will be used to compute control delay values
considering queue spillback and the results will be compared to the delay values
without queue spillback.
Exhibit 38-43 provides a schematic representation of the freeway facility in
the eastbound direction. Segments 3 (merge) and 5 (diverge) connect the freeway
to the urban street facility (Acadian Thruway).
Exhibit 38-43
Example Problem 2: Freeway
Facility Segmentation and O-D
Entry and Exit Points
The urban street facility consists of four signalized intersections and three
segments, as shown in Exhibit 38-44. The on-ramp terminal being analyzed is the
I-10 EB Ramps intersection.
Exhibit 38-44
Example Problem 2:
Urban Street Facility
Input Data
Signalized Intersection
The geometry of the intersection connected to the I-10 EB on-ramp is shown
in Exhibit 38-45. Three movements lead onto the on-ramp:
• NB right-turn (NBR): One channelized, unsignalized right-turn lane
• SB left-turn (SBL): One exclusive left turn lane with a protected phase
• EB through (EBT): One through lane
Exhibit 38-45
Example Problem 2A:
Signalized Intersection
Geometry: I-10 EB Ramps
Exhibit 38-46
Example Problem 2A: Phasing
Sequence: I-10 EB Ramps
The demand volumes for each analysis period are presented in Exhibit 38-47.
Additional input data are summarized in Exhibit 38-48.
Freeway Facility
The freeway facility (I-10 EB) is divided into seven segments, as shown in
Exhibit 38-49, where segment 3 (diverge) and segment 5 (merge) connect to the
subject signalized intersection (Acadian Thruway). Exhibit 38-50 summarizes the
facility’s geometric features.
Exhibit 38-49
Example Problem 2A:
Freeway Facility Segments
As shown, for a 120-s cycle, the capacity of the unsignalized NBR movement
is 34.8 vehicles. Aggregated to an hourly flow rate, the capacity is:
3,600
𝑐𝑁𝐵𝑅 = 34.8 × = 1,045 veh/h
120
Because of the actuated control operation, the discharge rate to the on-ramp
is different during each cycle and during each analysis period. Therefore, this
procedure must be repeated for every analysis period to compute the capacity of
the NBR unsignalized movement cNBR, as shown in Exhibit 38-52.
Exhibit 38-53 summarizes the calculations for this step. During analysis
period 3, the SBL movement operates over capacity (v/c = 1.56, highlighted in
red); therefore, its throughput to the ramp is constrained by its capacity of 685
veh/h. For all other movements and analysis periods, the throughput to the on-
ramp equals the demand because v/c < 1.
The calculated on-ramp demand is then provided as an input to the freeway
facility analysis (Exhibit 38-54). As shown, the ramp flow rates for the merge
segment (segment 5) are obtained from Exhibit 38-53 (highlighted in bold).
The results of the freeway facility analysis are provided in Exhibit 38-55.
Oversaturated conditions (LOS F) occur during analysis periods 2 and 3
(highlighted in red), therefore queueing may occur along the on-ramp.
Time Period 1 Time Period 2 Time Period 3 Time Period 4 Exhibit 38-54
Mainline Ramp Mainline Ramp Mainline Ramp Mainline Ramp Example Problem 2A: Freeway
Flow Flow Flow Flow Flow Flow Flow Flow Facility Demand Inputs
Segment Rate Rate Rate Rate Rate Rate Rate Rate
ID (veh/h) (veh/h) (veh/h) (veh/h) (veh/h) (veh/h) (veh/h) (veh/h)
1 5,209 — 6,300 — 5,300 — 5,000 —
2 5,209 348 6,300 450 5,300 1,200 5,000 50
3 4,861 135 5,850 116 4,100 1,000 4,950 96
4 4,726 — 5,734 — 3,100 — 4,854 —
5 4,726 975 5,734 1,203 3,100 1,411 4,854 567
6 5,701 — 6,937 — 4,511 — 5,421 —
7 5,701 — 6,937 — 4,511 — 5,421 —
The next step estimates the on-ramp queue length and compares the result to
the available queue storage length to determine whether spillback is expected to
occur. Exhibit 38-56 shows the expected on-ramp queues from the freeway
facility analysis. For each analysis period, the ramp storage ratio RQ is computed
by dividing the ramp queue by the available storage length (924 ft). During
analysis period 2, a queue is expected on the ramp, but it is not long enough to
cause queue spillback (RQ < 1). During analysis period 3, however, the on-ramp is
expected to have RQ = 2.31 (highlighted in red), which indicates that spillback will
occur at the intersection during this analysis period.
Exhibit 38-56 Analysis Ramp Flow Ramp Queue Ramp Ramp Storage Spillback
Example Problem 2A: Period Rate (veh/h) (veh) Queue (ft) Ratio (RQ) Expected?
Spillback Check: I-10 EB 1 975 0.0 0.0 0.00 No
On-Ramp 2 1,203 15.0 388.6 0.42 No
3 1,411 82.1 2,133.6 2.31 Yes
4 567 0.0 0.0 0.00 No
Because spillback will occur during at least one analysis period, its impacts
on the operation of the signalized intersection must be evaluated. The next
section illustrates the application of the methodology to evaluate spillback effects
at a signalized intersection.
Analysis Period 2
The procedure to evaluate queue spillback into intersections is applied for
analysis period 2, even though spillback is not expected to occur during this
analysis period. The application of the methodology is presented for this analysis
period to facilitate the understanding of the calculations.
Step 7A: Determine intersection throughput to on-ramp. The throughput
of movements into the on-ramp were determined previously as part of the queue
spillback check, as shown in Exhibit 38-53.
Step 7B: Obtain merging capacity with the freeway facilities method.
When the freeway facility operates in oversaturated conditions, the capacity of
the subject merge section may be constrained by the presence of queues along
the mainline. The Oversaturated Segment Evaluation procedure (Chapter 25)
computes the on-ramp queue ONRQ and on-ramp capacity ONRO every 15 s.
The merge capacity cmerge is then obtained by aggregating the ONRO parameter
into an hourly flow rate for each analysis period. Exhibit 38-57 shows the values
of ONRQ and ONRO over the study period (60 min), converted to hourly flow
rates.
Exhibit 38-57(a) compares the on-ramp capacity ONRO to the on-ramp
demand. During the first analysis period, undersaturated conditions exist along
the freeway, thus ONRO equals 2,000 pc/h, corresponding to the ramp roadway
capacity given in Exhibit 14-12, or 1,903 veh/h. During analysis periods 2 and 3,
oversaturated conditions occur and the on-ramp capacity drops to 5 pc per time
step, corresponding to 1,142 veh/h. During analysis period 4, the lower demand
along the freeway allows the mainline queue to clear within 4 time steps (60 s).
Therefore, during the first 60 s, the on-ramp capacity remains 1,142 veh/h. From
the fifth time step to the end of analysis period 4, there is no congestion at the
merge and therefore the on-ramp capacity is again 1,903 veh/h.
Exhibit 38-57
Example Problem 2A:
Freeway Segment 5 Merge
Capacity and Queue Lengths
Step 7C: Plot queue accumulation polygons (QAPs) for the on-ramp and
unsignalized movements. In this step, a QAP is plotted for the on-ramp as a
function of all protected and permitted movements entering the on-ramp, on a
cycle-by-cycle basis. Because an unsignalized movement (NBR) also discharges
into the on-ramp, a QAP must be developed for this movement as well. The
latter QAP is required to: (a) determine the discharge pattern of the unsignalized
movement throughout the cycle, and (b) allow the estimation of control delay for
this movement.
Exhibit 38-59 presents the QAPs for analysis period 2 for both the on-ramp
and the NBR movement.
Exhibit 38-59
Example Problem 2A:
Estimated Queue Lengths and
Merge Capacities During
Analysis Period 2
The cycle starts with a permitted left-turn movement (Φ1: SBL) discharging
into the on-ramp with a green time g1 = 43.9 s, divided in a queue service time gs1
= 40.2 s and a queue extension time ge1 = 3.7s (as defined in Chapter 31). During
the green interval for SBL, the capacity of the NBR movement is constrained
because drivers must yield to the protected left-turn vehicles. The saturation flow
rate for the NBR movement with a conflicting flow vSBL can be estimated from
Equation 38-27:
𝑣𝑜 𝑒 −𝑣𝑜 𝑡𝑐𝑔/3,600
𝑠𝑝 =
1 − 𝑒 −𝑣𝑜 𝑡𝑓ℎ/3,600
Because a queue is present in the NBR movement, the throughput for the
NBR movement is equal to its saturation flow rate:
𝜆𝑁𝐵𝑅,2 = 𝑠𝑁𝐵𝑅,𝑝𝑒𝑟𝑚2 = 1,282 veh/h
where λNBR,2 is the throughput for the NBR movement during the SBL green
extension (veh/h/ln) and sNBR,perm2 is the saturation flow rate of the NBR
movement during the SBL green extension time (veh/h/ln).
With the discharge patterns for the NBR determined, the on-ramp’s queue
profile during Φ1 can be determined. During the SBL queue service time (cycle
time t = 0 to t = 40.2 s), the throughput to the on-ramp is given by:
𝜆𝑂𝑁𝑅 = 𝜆𝑆𝐵𝐿 + 𝜆𝑁𝐵𝑅,1 = 1,739 + 282 = 2,021 veh/h = 0.561 veh/s
Given that the merge capacity cmerge is 1,142 veh/h for the current analysis
period, the on-ramp queue will grow at the following rate during the SBL queue
service time:
𝜆𝑂𝑁𝑅 − 𝑐merge = 2,021 − 1,142 = 879 veh/h = 0.244 veh/s
Therefore, at the end of the SBL queue service time (t = 40.2s), the queue at
the on-ramp will be 0.244 x 40.2 = 9.8 vehicles (Exhibit 38-59a).
This process is then repeated for all phases throughout the cycle. The results
for a single cycle (120 s) are presented in Exhibit 38-60, where the maximum on-
ramp queue occurs at t = 50.48 s, with 10.82 veh. The expected on-ramp queue at
the end of the cycle is 2.02 veh. The remaining cycles within analysis period 2
show the same pattern, where the on-ramp queue at the end of each cycle
becomes the initial queue at the start of the next cycle.
Each row in Exhibit 38-60 describes a portion of the cycle, as follows:
• gs1 is the queue service time for SBL (Φ1), as previously discussed.
• ge1 is the green extension time for SBL (Φ1). The NBR movement
discharges at the permitted saturation flow rate due to the queue that has
developed during gs1 and the on-ramp queue grows at a rate of 0.07 veh/s.
• r1 is the effective red time for SBL (Φ1). There is no throughput from
protected movements and the NBR movement discharges freely at the
saturation flow rate. The on-ramp queue grows at a rate of 0.11 veh/s.
• g2* is the effective green for NBT (Φ2), with no throughput from protected
movements. The duration of 0.88 s is calculated based on the queue
service time of the NBR approach. The on-ramp queue grows at a rate of
0.11 veh/s.
• g2** is the remaining effective green for NBT (Φ2). For this portion, no
queue remains on the NBR approach, therefore the NBR throughput is
equal to its demand flow rate vNBR. The on-ramp queue discharges at a
rate of 0.17 veh/s.
• r2 is the effective red time for NBT (Φ2). There is no throughput from
protected movements and the NBR throughput is equal to its demand
flow rate vNBR. The on-ramp queue discharges at a rate of 0.17 veh/s.
• gs7 is the queue service time for EBT (Φ7). The EBT discharges into the on-
ramp at the saturation flow rate. The throughput of the NBR movement is
restricted to the permitted saturation flow rate, causing queues to develop
in the NBR approach. The on-ramp queue grows at a rate of 0.26 veh/s.
• ge7* is the green extension time for EBT (Φ7). The duration of 0.03 s is
calculated based on the queue service time of the NBR approach. The
NBR movement discharges at the permitted saturation flow rate. The on-
ramp queue grows at a rate of 0.08 veh/s.
• ge7** is the remaining extension time for EBT (Φ7). The EBT movement
discharges at a rate equal to its arrival flow rate during the effective
green. For this portion, no queue remains on the NBR approach, therefore
the NBR throughput is equal to its demand flow rate vNBR. The on-ramp
queue discharges at a rate of 0.15 veh/s.
• r7 is the effective red time for EBT (Φ7). There is no throughput from
protected movements and the NBR throughput is equal to its demand
flow rate vNBR. The on-ramp queue discharges at a rate of 0.17 veh/s.
At the end of the analysis period, a residual queue of 23.32 veh is expected
along the on-ramp, and this value is carried to the start of the next analysis
period. The analysis period length of 900 s does not correspond to an exact
number of signal cycles, and the last cycle is interrupted at t = 60 s. Therefore, the
next analysis period will start the analysis from the same timestamp to maintain
consistency.
Step 7D: Calculate equivalent capacities for the affected movements.
Because spillback does not occur during analysis period 2, no adjustment to the
intersection capacity is necessary.
Analysis Period 3
The same analysis steps performed for analysis period 2 are applied again
for analysis period 3.
Exhibit 38-61
Example Problem 2A:
Estimated Queue Lengths and
Merge Capacities During
Analysis Period 3
Queue spillback occurs during the third cycle (SBL queue service time),
when the on-ramp queue reaches the maximum storage LONR of 35.5 veh. At this
time, the maximum flow rate that can enter the on-ramp is constrained by the
merge capacity cmerge. In other words, the maximum number of vehicles allowed
to enter the ramp is equal to the number of vehicles that are able to merge to the
freeway mainline. In addition, the queues developed in the NBR are longer
during cycles 3 through 8, causing an increased delay for this movement due to
the queue spillback conditions at the on-ramp.
The on-ramp queue at the start of cycle 3 is 27.9 veh. The cycle starts with the
SBL movement, with an effective green time g1 of 47.3 s. Because this movement
already operates with v/c > 1, the queue service time gs1 is equal to g1, and no
green extension time is available (ge1 = 0). The protected movement then
discharges at a saturation flow rate sSBL of 0.483 veh/s, while the NBR movement
discharges at a permitted saturation flow rate sNBR of 0.078 veh/s. At the same
time, the on-ramp discharges to the freeway at a rate cmerge of 1,142 veh/h, equal
to 0.317 veh/s. Therefore, the on-ramp queue grows at the following rate:
𝜆𝑂𝑁𝑅 − 𝑐merge = (0.483 + 0.078) − 0.317 = 0.244 veh/s
The time remaining until spillback occurs is calculated by dividing the
remaining on-ramp queue storage by the queue growth rate:
35.5 − 27.9
Time to spillback = = 31.2 s
0.244
Spillback is expected to occur within 31.2 s of the onset of g1. The total
effective green g1 value of 47.3 s is then divided into two portions:
• gs1* (31.2 s), discharging at the saturation flow rate; and
• gs1,sp (16.1s), the remaining time that is affected by queue spillback,
limiting the maximum discharge to the on-ramp to the merge capacity
cmerge of 0.317 veh/s. Note that this constraint is shared by two movements
entering the on-ramp (SBL and NBR).
The effect of queue spillback on the intersection capacity during gs1,sp is then
measured by the capacity reduction factor β1,sp, defined as the ratio between the
maximum on-ramp capacity during queue spillback and the throughput from
the intersection movements (SBL and NBR):
𝑐merge 0.317
𝛽1,𝑠𝑝 = = = 0.565
𝜆𝑆𝐵𝐿 + 𝜆𝑁𝐵𝑅 (0.483 + 0.078)
A capacity reduction of 0.565 means that only 56.5% of the expected
intersection throughput is able to enter the on-ramp when queue spillback occurs
during phase gs1,sp. The capacity adjustment factor is applied to each movement
to obtain their adjusted throughputs for this analysis period:
𝜆𝑆𝐵𝐿,𝑎𝑑𝑗 = 𝜆𝑆𝐵𝐿 × 𝛽1,𝑠𝑝 = 0.483 × 0.565 = 0.273 veh/s
𝜆𝑁𝐵𝑅,𝑎𝑑𝑗 = 𝜆𝑁𝐵𝑅 × 𝛽1,𝑠𝑝 = 0.078 × 0.565 = 0.044 veh/s
The procedure is then repeated for the remaining movements of the cycle, as
shown in Exhibit 38-62.
Analysis Period 4
The same steps performed for analysis periods 2 and 3 are applied again in
analysis period 4.
Step 7A: Determine intersection throughput to on-ramp. The throughput
for movements that enter the on-ramp was previously determined as part of the
queue spillback check and shown in Exhibit 38-53.
Step 7B: Obtain merging capacity with the freeway facility method. The
merge capacity for analysis period 4 was previously determined, as shown in
Exhibit 38-57a. Because the congestion along the freeway mainline is dissipating
during this analysis period, the merge capacity is not constant: from time steps 1
through 4, the merge capacity is 1,142 veh/h, consistent with oversaturated
conditions from previous time periods. After time step 5, the merge capacity is
equal to the ramp roadway capacity (1,904 veh/h).
Step 7C: Plot QAPs for the on-ramp and unsignalized movements. The
procedure described earlier is applied to plot the QAPs, shown in Exhibit 38-64.
Queue spillback occurs during the first cycle, due to the residual queue from the
previous time period. However, due to low volumes at the intersection and
improvement of performance along the freeway mainline, the on-ramp clears
quickly. The queue has cleared by the end of the second cycle.
Exhibit 38-64
Example Problem 2A:
Estimated Queue Lengths and
Merge Capacities During
Analysis Period 4
Step 7D: Calculate adjusted capacities for the affected movements. The
procedure described earlier is used to calculate the capacity reduction factor for
the SBL movement, as shown in Exhibit 38-65. The estimated capacity reduction
is minor, as spillback only occurs during the first cycle. The EBT movement does
not experience queue spillback, therefore no adjustment is necessary.
Time SBL Movement capacity (veh/h) SBL Control delay (s/veh) Exhibit 38-66
Period Without spillback With spillback Without spillback With spillback Example Problem 2A:
1 652 652 60.3 60.3 Performance Measure
2 586 586 55.9 55.9 Comparison with and without
3 685 482 293.5 589.2 Consideration of Spillback
4 746 721 575.2 609.5 Effects
Exhibit 38-67
Example Problem 2B: TWSC
Intersection Geometry: I-10
EB Ramps
The on-ramp demand estimates are then used as inputs to the freeway
facility analysis. In this example, spillback will occur in analysis period 3.
Step 9B: Obtain Merging Capacity Using the Freeway Facility Methodology
This step computes the merging capacity into the freeway cmerge. The freeway
facility inputs are obtained from Exhibit 38-68, yielding the following results:
• Analysis period 2: No queue spillback.
• Analysis period 3: 1,142 veh/h.
• Analysis period 4: 1,142 veh/h during 15 time steps (222 s), and then 1,903
veh/h. This analysis period considers a lower merge capacity while a
mainline queue is present during the first 222 s. For the remainder of the
analysis period, the merge capacity is constrained only by the on-ramp
capacity, similar to the scenario presented in Example Problem 2A.
The results show that queue spillback occurs only during analysis period 3.
The initial queue of analysis period 3 is 0 and it takes 3.8 min for the on-ramp to
reach its maximum storage capacity. Therefore, the spillback time Tsp is computed
as 15 – 3.8 = 11.2 min. Exhibit 38-70 depicts the on-ramp’s QAP, based on the
table results.
Exhibit 38-70
Example Problem 2B: Queue
Accumulation Polygon for the
On-Ramp
𝑐merge × 𝑣𝐸𝐵𝑇 × 96
𝑐𝑠𝑝,𝐸𝐵𝑇 = = = 61 veh/h
𝑣𝑆𝐵𝐿 + 𝑣𝑁𝐵𝑅 + 𝑣𝐸𝐵𝑇 1,071 + 630 + 96
The equivalent capacities cEQ,i for each movement i, aggregated for the 15-
min analysis period, are obtained proportionately to the spillback time Tsp:
Exhibit 38-72
Example Problem 2C: AWSC
Intersection Geometry: I-10
EB Ramps
The estimated on-ramp demand values are provided as inputs to the freeway
facility analysis. The freeway facility is then analyzed and the expected on-ramp
queues are determined as shown in Exhibit 38-74.
Because spillback will occur during analysis period 3, its impacts on the
intersection’s operation must be evaluated. The next section illustrates the
application of the queue spillback evaluation methodology at an AWSC
intersection.
Step 13B: Obtain Merging Capacity with the Freeway Facilities Method
In this example, the ramp metering rate (900 veh/h) is an additional input to
the freeway facility analysis and acts as a constraint to the merge capacity.
Therefore, the merge capacity for this analysis is kept constant at 900 veh/h.
Exhibit 38-76 illustrates the QAP for the on-ramp, based on the results shown
in Exhibit 38-75.
Exhibit 38-76
Example Problem 2C: Queue
Accumulation Polygon for the
On-Ramp
Capacity (veh/h) Control Delay (s/veh) Departure Headway (s) Exhibit 38-78
Demand Without With Without With Without With Example Problem 2C:
Movement (veh/h) Spillback Spillback Spillback Spillback Spillback Spillback Performance Measure
Comparison with and without
EBL 75 359 359 15.6 15.6 10.0 10.0
Consideration of Spillback
EBT 19 396 212 12.6 21.7 9.1 17.0
Effects—Analysis Period 3
NBT 229 497 497 16.3 16.3 7.2 7.2
NBR 539 550 497 58.9 92.3 6.5 7.3
SBL 546 462 454 128.0 136.5 7.8 7.9
SBT 220 494 494 16.0 16.0 7.3 7.3
Note: EBL = eastbound left turn, EBT = eastbound through, NBT = northbound through, NBR = northbound right
turn, SBL = southbound left turn, SBT = southbound through.
Exhibit 38-79
Example Problem 3:
Study Site
Exhibit 38-80
Example Problem 3:
Freeway Facility Geometry
Input Data
Traffic demands for the freeway facilities and ramps are provided in Exhibit
38-81 for each 15-min analysis period.
Exhibit 38-81 Freeway Facility 1 (I-75 SB) Freeway Facility 2 (SR-826 SB)
Example Problem 3: Mainline Demand Diverge Demand Mainline Demand Merge Demand
Traffic Demands Analysis Flow Rate Flow Rate Flow Rate Flow Rate
Period (veh/h) (veh/h) (veh/h) (veh/h)
1 5,400 1,400 4,000 1,400
2 6,200 3,000 5,700 3,000
3 6,000 3,400 5,600 3,400
4 4,500 800 4,500 800
Spillback Check
Analyzing SR-826 using Chapter 25’s Freeway Facilities Oversaturated
Segment Evaluation methodology provides the expected on-ramp queue for
every analysis period. The first check compares the off-ramp demand to the
ramp roadway capacity, as shown in Exhibit 38-84. The ramp queue starts to
develop during analysis period 2. At the end of this time period, a ramp queue
length of 1,188 ft is expected, yielding a queue storage ratio of 0.33. Therefore,
spillback onto I-75 is not expected during analysis period 2. However, during
analysis period 3, a ramp queue length of 5,160 ft is expected, yielding a queue
storage ratio of 1.41. Therefore, spillback onto I-75 will occur during analysis
period 3.
Spillback Analysis
Because spillback is expected to occur, the methodology described in
Appendix A (Exhibit 38-A8) is applied to evaluate how it affects I-75 SB. The
methodology’s application to each analysis period is presented below.
Analysis Period 1
No oversaturated conditions occur; therefore, no additional calculations are
needed for this analysis period.
Analysis Period 2
During analysis period 2, the downstream merge segment operates at LOS F
and the on-ramp capacity is expected to be reduced.
Exhibit 38-85
Example Problem 3:
Link–Node Structure for
Spillback Analysis: I-75 SB
period, the merge capacity is constant at 13.4 pc/ts or 3,217 pc/h, while the ramp
demand is 14.0 pc/ts or 3,369 pc/h.
Given the merge’s demand and capacity, the queue in the ramp roadway
increases by 0.6 pc during every time step. Exhibit 38-86 illustrates the ramp
queue and the total number of vehicles in the ramp, considering an initial value
of 50.8 pc in the ramp at the start of the analysis period, as previously computed.
Exhibit 38-86
Example Problem 3: Queued
Vehicles and Total Number of
Vehicles RNV in the Ramp:
Analysis Period 2
Exhibit 38-87
Example Problem 3: Ramp
Capacity RSTG and Ramp
Input RI: Analysis Period 2
Because spillback does not occur, no additional calculations for the mainline
are required.
𝑅𝐹(𝑖, 𝑝, 𝑘) 1,679.5
𝑆𝑅(𝑖, 𝑝, 𝑘) = = = 31.9 mi/h
𝑅𝐾(𝑖, 𝑝, 𝑘) 71.6
Analysis Period 3
Steps 1, 2, and 9: Ramp Analysis
The same steps are repeated for analysis period 3. The ramp analysis is
summarized in Exhibit 38-88. During this analysis period, the ramp demand is
15.4 pc/ts, while the merge capacity is 13.9 pc/ts. Because demand is greater than
capacity, the number of vehicles increases gradually, causing the capacity
constraint RSTG to decrease each time step. At time step 14, the value of RSTG
becomes equal to the merge capacity (13.9 pc/ts), which implies that the ramp has
reached jam density and the maximum flow that can enter the ramp is equal to
the flow departing the ramp. Therefore, queue spillback into the mainline starts
at time step 15.
Exhibit 38-88
Example Problem 3: Ramp
Capacity RSTG and Ramp
Input RI: Analysis Period 3
After the onset of queue spillback, the number of unserved vehicles at the
exit is computed every time step through the parameter OFRUV(i, t, p). Then, the
expected length of the mainline queue OFRLQ(i, t, p) is computed based on the
number of unserved vehicles and the ramp queue density RKQ, as given by
Equation 38-A35:
𝑂𝐹𝑅𝑈𝑉(𝑖, 𝑡, 𝑝)
𝑂𝐹𝑅𝐿𝑄(𝑖, 𝑡, 𝑝) =
𝑅𝐾𝑄 (𝑖, 𝑡, 𝑝)
Exhibit 38-89
Example Problem 3: Spillback
Queue Length on I-75 SB:
Analysis Period 3
The parameter OFRLQ represents the length of the queue if all unserved
vehicles were queued in a single line. Given the segment geometry (Exhibit 38-
90), the operating regimes and flow modes can be obtained as a function of
OFRLQ:
• 0 < OFRLQ ≤ 1,400 ft: Regime 1
• 1,400 ft < OFRLQ ≤ 3,000 ft: Regime 4, with increased turbulence
• 3,000 ft < OFRLQ: Regime 4, with lane blockage (queue extends upstream
beyond the diverge)
Exhibit 38-90
Example Problem 3: Available
Queue Storage on I-75 SB
Exhibit 38-91
Example Problem 3: Back of
Queue Length, Including
Queue Influence Area, at the
End of Analysis Period 3
𝑅𝐹(𝑖, 𝑝, 𝑘) 1,707
𝑆𝑅(𝑖, 𝑝, 𝑘) = = = 21.5 mi/h
𝑅𝐾(𝑖, 𝑝, 𝑘) 108.4
Performance measures are computed for the blocked and unblocked portions
of each segment of the freeway facility.
Segment 1-3 (diverge)—blocked portion. Similar to the ramp, the flow
through the blocked portion is aggregated for this time period:
𝑆
Analysis Period 4
During analysis period 4, the congestion at the downstream facility (SR-826)
dissipates, which allows the ramp to discharge at the ramp roadway capacity
(4,400 pc/h, or 18.33 pc/ts). Given the low ramp demand during this time period,
the queue clears quickly (9 time steps, or 135 s). After the 10th time step, the
freeway facility returns to undersaturated conditions.
Input Data
The site layout is shown in Exhibit 38-92. The location’s traffic and geometric
characteristics are as follows:
• Single-lane approaches on all roundabout entries
• Adjusted demand flow rates in pc/h for all movements are as shown in
Exhibit 38-92
• No heavy vehicles
• U-turn movements are negligible
• Pedestrian activity limits the exit capacity to the on-ramp to 1,300 pc/h.
• Ramp length = 1,657 ft
• The on-ramp connecting the roundabout to the freeway is metered at a
rate cRM = 800 pc/h.
Exhibit 38-92
Example Problem 4:
Study Interchange Schematic
3,600
(1,380) 0.22 1,380
𝑄95,𝑆𝐵 √ 2
= 900(0.25) 0.22 − 1 + (1 − 0.22) + ( ) = 1 veh
150(0.25) 3,600
[ ]
Similarly,
𝑄95,𝐸𝐵 = 6 veh
𝑄95,𝑁𝐵 = 121 veh
These values are rounded to the nearest vehicle.
Exhibit 38-93 provides the flows and resulting queues at the roundabout.
Exhibit 38-94
Example Problem 4:
Roundabout Approach Priority
Order
Starting with the Rank 1 (SB) approach, the maximum throughput for the
movement exiting through the EB leg (the on-ramp) is calculated as follows:
3,600
Equation 38-28 𝜆𝑆𝐵𝐿,𝑝𝑐𝑒 = min (𝑣𝑆𝐵𝐿,𝑝𝑐𝑒 , 𝑐𝑝𝑐𝑒,𝑆𝐵 × 𝑝𝑆𝐵𝐿 , )
ℎ𝑠
where
λSBL,pce = maximum throughput for the SB left-turn movement (pc/h),
vSBL,pce = flow rate for the SB left-turn movement (pc/h),
cpce,SB = entry lane capacity for the SB roundabout approach (pc/h),
pSBL = percent of demand from the SB approach into the on-ramp
= vSBL,pce divided by the total flow rate for the SB approach, and
hs = departure saturation headway into the on-ramp (s/veh).
Then:
100 3,600
𝜆𝑆𝐵𝐿,𝑝𝑐𝑒 = min (100, [1,380 × ], ) = 100 pc/h
300 2.77
Because the Rank 3 (NB) approach is the only one with a volume-to-capacity
ratio over 1, the conflicting flows and capacity values calculated above are valid.
The next calculation is the maximum throughput for the remaining
movements of the approach that contribute to the conflicting flows for
downstream approaches. This calculation is performed as follows:
𝜆𝑆𝐵𝑇,𝑝𝑐𝑒 = min(𝑣𝑆𝐵𝑇,𝑝𝑐𝑒 , 𝑐𝑝𝑐𝑒,𝑆𝐵 × 𝑝𝑆𝐵𝑇 ) Equation 38-29
where
λSBT,pce = maximum throughput for the southbound-through movement (pc/h),
vSBT,pce = flow rate for the southbound-through movement (pc/h),
cpce,SB = entry lane capacity for the southbound roundabout approach (pc/h), and
pSBL = percent of demand from SB approach for through movement
= vSBT,pce divided by the total flow rate for the SB approach.
Then:
200
𝜆𝑆𝐵𝑇,𝑝𝑐𝑒 = min (200, [1,380 × ]) = 200 pc/h
300
The maximum throughput for each approach and O-D pair is calculated
considering the maximum throughput on the on-ramp and accounting for
higher-rank approaches:
3,600
𝜆𝐸𝐵𝑇,𝑝𝑐𝑒 = min (𝑣𝐸𝐵𝑇,𝑝𝑐𝑒 , 𝑐𝑝𝑐𝑒,𝐸𝐵 × 𝑝𝐸𝐵𝑇 , − 𝜆𝑆𝐵𝐿 )
ℎ𝑠
500 3,600
𝜆𝐸𝐵𝑇,𝑝𝑐𝑒 = min (500, [1,016 × ],[ − 100]) = 500 pc/h
700 2.77
The maximum throughput to the on-ramp is lower than the exit capacity
(1,300 pc/h), thus the northbound approach flow rate is limited by its own
approach capacity.
3,600
3,600 × 0.22
√ 1,380
𝑑𝑆𝐵 = + 900(0.25) 0.22 − 1 + (0.22 − 1)2 +
1,380 450(0.25)
[ ]
+ 5 × min[0.22,1]
2 3,600 2,100
3,600 2,100 2,100 ×
𝑑𝑠𝑝 = + 900(0.25) [ √
−1+ ( − 1) + 800 800 ]
800 800 800 450(0.25)
2,100
+ 5 × min [ , 1]
800
𝑑𝑠𝑝 = 747.94 s/veh
Therefore, the total average delay per approach is:
100
𝑑𝑆𝐵,𝑇 = 𝑑𝑆𝐵 + 𝑑𝑠𝑝 × = 65.05 s/veh
1,234
500
𝑑𝐸𝐵.𝐿 = 𝑑𝐸𝐵 + 𝑑𝑠𝑝 × = 317.54 s/veh
1,234
634
𝑑𝑁𝐵,𝑅 = 𝑑𝑁𝐵 + 𝑑𝑠𝑝 × = 1,020.14 s/veh
1,234
5. REFERENCE
This reference can be found in 1. University of Florida Transportation Institute; Cambridge Systematics, Inc.;
the Technical Reference
Library in Volume 4. and A. Skabardonis. NCHRP Web-Only Document 290: Highway Capacity
Manual Methodologies for Corridors Involving Freeways and Surface Streets.
Transportation Research Board, Washington, D.C., 2020.
Exhibit 38-A1
Off-Ramp Queue Spillback
Check Flowchart
CAPACITY CHECKS
The procedure first determines whether capacity is exceeded at any of the
critical points along the diverge section.
Exhibit 38-A2 Ramp FFS, SFR (mi/h) Single-Lane Ramps Two-Lane Ramps
Capacity of Ramp Roadways >50 2,200 4,400
(pc/h) >40–50 2,100 4,200
>30–40 2,000 4,000
≥20–30 1,900 3,800
<20 1,800 3,600
Notes: Capacity of a ramp roadway does not ensure an equal capacity at its freeway or other high-speed junction.
Junction capacity must be checked against criteria in Exhibit 14-10 and Exhibit 14-11.
FFS = free-flow speed.
where
Qg,i = queue growth during analysis period i (veh),
vR = off‐ramp demand (pc/h),
cR = off-ramp roadway capacity (pc/h),
fHV = adjustment factor for heavy vehicle presence,
PHF = peak hour factor, and
T = analysis time period (0.25 h for a 15-min period) (h).
The ramp queue during the first time period of the analysis must be zero,
otherwise the analysis’ time–space domain boundaries need to be re-evaluated.
The accumulated queue length at the end of analysis period t is the cumulative
value of Qg,i until t:
𝑡
The maximum queue length Qmax during the entire study period is the
maximum value of Qt obtained using Equation 38-A2 and is used as an input for
the next stage of the spillback check procedure.
Signalized Intersections
The methodology of Chapters 19 and 31 evaluates the performance of
individual lane groups for a subject approach. It also estimates the back-of-queue
length Q (Equation 31-149) or a percentile back-of-queue length Q% (Equation
31-150). In some cases, only one high-demand movement on the intersection
approach is the bottleneck that results in spillback, yielding an unbalanced lane
usage pattern at the ramp. Field observations have shown that urban street
intersection failures may occur for one lane group. As drivers position themselves
in a specific lane at the ramp in anticipation of the downstream signal, the ramp’s
lane usage becomes unbalanced, as illustrated in Equation 38-A3.
Exhibit 38-A3
Examples of Unbalanced
Ramp Lane Usage
When off-ramps have two or more lanes, the estimated queue length for each
intersection lane group must be associated with specific ramp lanes. Exhibit 38-
A4 illustrates an example of a typical ramp terminal for a two-lane off-ramp.
Drivers that desire to make a left turn at the intersection will position themselves
in the leftmost lane (ramp lane 2), while drivers who intend to turn right will
likely choose the rightmost lane (ramp lane 1). Analyst judgement is required to
define the grouping of intersection lane groups into ramp lanes.
Exhibit 38-A4
Illustrative Assignment of
Intersection Lane Groups to
Ramp Lanes
where
Ql,k = number of queued vehicles in ramp lane k during a 15-min interval
(veh);
QLG,m = number of queued vehicles from lane group m associated with ramp
lane k during a 15-min interval (veh);
Q%,LGn = estimated back of queue length (nth percentile), from Equation 38-A5
(derived from Equation 31-150) (veh/ln); and
NLG.m = number of approaching lanes for lane group m.
with
𝑄%,𝐿𝐺𝑛 = (𝑄1 + 𝑄2 )𝑓𝐵% + 𝑄3 Equation 38-A5
where
𝑄𝑖 = ith-term back-of-queue size (veh/ln), from Equation 31-141 to Equation
31-143; and
𝑓𝐵% = percentile back-of-queue factor corresponding to the nth percentile,
from Equation 31-151 or 31-153.
Unsignalized Intersections
Each unsignalized intersection type has its own methodology to estimate
queue length. The TWSC methodology estimates the 95th percentile queue
length for minor movements with Equation 20-68, while the 95th percentile
queue length for AWSC approaches is estimated with Equation 21-33. For
roundabouts, the 95th percentile queue length for a given lane is provided by
Equation 22-20. Once the lane group queue(s) have been determined, they are
assigned to ramp lanes following the same procedure described above for
signalized intersections.
𝐿𝑅 = ∑ 𝑁𝑖 × 𝐿𝑖 Equation 38-A8
𝑖
where
LR = ramp storage length (ft),
Ni = number of lanes in section i, and
Li = length of section 𝑖 (ft).
The individual ramp storage for each of the k lanes in the off-ramp LR.k can be
estimated by assigning the intersection lane groups to ramp lanes, as previously
described:
Off-Ramp Operations
To evaluate the interaction between the freeway mainline and the
downstream off-ramp terminal, the link–node approach used by Chapter 25 to
evaluate oversaturated freeway segments is expanded, with additional links and
nodes used to represent the off-ramp segment. As shown in Exhibit 38-A5, the
mainline node for the off-ramp (node 3) is connected to the off-ramp segment,
which has a three-node structure:
• Ramp node 3.1: Interface between the freeway diverge segment (exit lanes)
and the upstream end of the ramp roadway. The volume that flows
through this node is equivalent to the number of vehicles that are able to
leave the freeway.
• Ramp node 3.2: Interface between the ramp roadway and the arterial
intersection approach. The volume that flows through this node is
equivalent to the number of vehicles that are able to leave the ramp
roadway and approach the intersection;
• Ramp node 3.3: Discharge capacity of the arterial intersection approach.
The volume that flows through this node is equivalent to the number of
vehicles that are able to enter the intersection.
Exhibit 38-A5
Expanded Link–Node
Structure to Evaluate Off-
Ramp Segments
defined by the point where additional lanes are provided. When modeling the
off-ramp geometry, the method considers the channelization at the intersection
approach, because imbalances in the turning movement demands may cause
queues on a subset of lanes. Exhibit 38-A6 shows a typical queue formation
resulting from a left-turn movement that operates with insufficient capacity. In
this scenario, the approaching left-turn vehicles are positioned in the leftmost
lane and spillback may occur even when some approach lanes are undersaturated.
Exhibit 38-A6
Example Off-Ramp Geometry
with Heavy Left-Turn Demand
at a Signalized Intersection
The type of ramp terminal is an important input into the analysis. Signalized
intersections operate in cyclical patterns, and therefore have fluctuating queue
lengths. For certain demand scenarios, this pattern can result in queues backing
up into the freeway and then discharging multiples times within a 15-min
analysis period.
STOP-controlled intersections and downstream merge segments (in the case
of a freeway-to-freeway connection) have more uniform discharging rates. For
cases other than signalized intersections, off-ramp queues are assumed to develop
or discharge linearly based on the relationship between demand and capacity.
Spillback Regimes
The impact of queue spillback on the freeway mainline varies as a function of
the queue length and the lanes blocked. Five spillback regimes are defined (A-1)
and are illustrated in Exhibit 38-A7.
Regime 0
Under this regime, shown in Exhibit 38-2(a), there are no queues in the ramp
roadway or the queue, if it exists, is contained within the ramp roadway
boundaries. There are no operational effects in the ramp influence area.
Regime 1
The queue ends within the deceleration lane and does not spill back into the
freeway mainline, as shown in Exhibit 38-A7(b). During undersaturated
conditions, the deceleration lane serves as a transition zone between speeds on
the mainline (typically 55–75 mi/h) and the advisory speed posted for the off-
ramp (typically 20–50 mi/h). When queues begin to form on the deceleration
lane, the available deceleration distance is reduced and speeds along the
rightmost lane are affected.
Exhibit 38-A7
Off-Ramp Queue Spillback
Regimes
(b) Regime 1: Queue at the deceleration lane (c) Regime 2: Queue along the shoulder
(d) Regime 3: Queue along the rightmost lane (e) Regime 4: Queue blockage of the adjacent lane
Regime 2
The queue extends upstream beyond the deceleration lane, but sufficient
lateral clearance on the right-hand shoulder provides additional queue storage.
As shown in Exhibit 38-A7(c), there is no transition zone within the deceleration
lane. Drivers decelerate and join the back of the queue more abruptly, resulting
in turbulence and reduced speeds in the rightmost lane. If no lateral clearance
exists immediately upstream of the deceleration lane, Regime 2 conditions are
not possible. In some cases, this regime does not occur even when storage is
available, depending on site-specific driver behavior.
Regime 3
The queue extends to the rightmost freeway mainline lane, as shown in
Exhibit 38-A7(d). This regime may occur when no shoulder is available for
additional queue storage or when drivers choose to queue in the rightmost lane
once the deceleration lane is entirely occupied. Non-exiting vehicles on the
rightmost lane are delayed or change lanes, which causes increased turbulence
and reduced speeds in the two rightmost lanes.
Regime 4
The queue blocks the rightmost lane, and drivers occasionally or often use
the next freeway mainline lane to the left to force their way into the queue, thus
blocking an additional lane, as shown in Exhibit 38-A7(e). During this regime,
freeway speed and capacity are significantly reduced. The effects of spillback
vary by site and time interval due to differences in driver behavior and site
geometry. Data collection at locations around the United States has shown that
drivers block the adjacent lane at some sites, but do not at other sites, regardless
of the queue spillback length at a given site.
Facility Variable
• QIA(i, p)—Length of the queue influence area (ft) for segment i during
analysis period p, measured from the back of the queue.
Segment Variables
• ΔK(i, p)— additional density in the queued mainline lines in segment i
during analysis period p (pc/mi/ln).
• ΔNV (i, t, p) — additional number of passenger cars in the congested
portion of segment i due to an off-ramp queue during time step t in
analysis period p (pc),
• KBBL(i, j)—background density (pc/mi/ln) at the blocked lanes in segment
i, when queue spillback occurs at a downstream segment j.
• KBUB(i, j)—background density (pc/mi/ln) at the unblocked lanes in
segment i, when queue spillback occurs at a downstream segment j.
• KQBL(i, t, p)— queue density (pc/mi/ln) of the blocked portion of segment
i during time step t in analysis period p.
• KQUB(i, t, p)— queue density in the unblocked portion of segment i
during time step t in analysis period p (pc/mi/ln).
• L(i) —length of segment i (ft).
• LCR(i, t, p)—rate of lane change maneuvers in the queue influence area
upstream of a queue from an off-ramp, for segment i during time step t in
analysis period p.
• LD(i, p)—available deceleration lane length (ft) for segment i during
analysis period p. This variable is used to calculate performance measures
for ramp segments.
• MO2UB(i, t, p) —maximum number of passenger cars that can enter the
unblocked portion of segment i during time step t in analysis period p due
to the presence of a queue in the downstream ramp segment.
• MO2BL(i, t, p) —maximum number of passenger cars that can enter the
blocked portion of segment i during time step t in analysis period p due to
the presence of a queue in the downstream ramp segment.
• MQ1(i, t, p)—queue length of off-ramp unserved vehicles in the rightmost
mainline lane for segment i during time step t in analysis period p.
• MQ2(i, t, p)—queue length of off-ramp unserved vehicles in the second-to-
the-right mainline lane, for segment i during time step t in analysis period
p. If Regime 4 is not expected to occur, this parameter value is set to zero.
• NQ(i)—number of blocked lanes if the off-ramp queue backs up into the
freeway mainline. This parameter is a function of the prevailing spillback
regime at segment i as provided by the analyst. The value for this
Node Variables
• CAFBL(i, t, p)—capacity adjustment when one or more lanes of segment i
are entirely blocked during time step t in analysis period p. This variable
is used to calculate friction effects that cause through vehicles to slow
down due to the presence of a queue in the rightmost lanes.
• CAFUP(i, t, p)—capacity adjustment factor for node i during time step t in
analysis period p. This variable affects approaching vehicles within the
queue influence area (QIA) upstream of an off-ramp queue. It accounts for
the turbulence caused by intense lane changing within the QIA as vehicles
adjust their position when there is a downstream off-ramp queue.
• MFBL(i, t, p)—mainline flow rate that can cross the blocked portion of
node i during time step t in analysis period p.
• MFUB(i, t, p)—mainline flow rate that can cross the unblocked portion of
node i during time step t in analysis period p.
• MIBL(i, t, p)—maximum flow desiring to enter the blocked portion of
node i during time step t in analysis period p.
Ramp Variables
• RC(i, t, p)—capacity of the ramp roadway (pc/ts) from segment i during
time step t in analysis period p. Obtained by dividing the capacity values
for the ramp roadway (pc/h) provided in Exhibit 14-12 in Chapter 14 by
the number of time steps in one hour (240).
• RF(i, t, p, k)—flow (pc/ts) that can enter the ramp roadway at segment i
during time step t in analysis period p using branch k.
• RI(i, t, p, k)—maximum flow (pc/ts) desiring to enter the off-ramp on
segment i during time step t in analysis period p using branch k, including
queues accumulated from previous time periods.
• RKC—ramp density at capacity (pc/mi/ln).
• RKQ(i, t, p, k)—ramp roadway queue density (pc/mi/ln) for segment i
during time step t in analysis period p using branch k.
• RL(i)—ramp roadway length (ft) for segment i.
• RN(i, p, k)—number of ramp lanes for branch k of segment i in analysis
period p. Similar to the number of mainline lanes, it could vary by time
interval if a temporary lane closure is in effect.
• RNV(i, t, p, k)—maximum number of passenger cars within the ramp of
segment i at the end of time step t during analysis period p using branch
k. The number of passenger cars is based initially on the calculations of
Chapters 12, 13, and 14, but as queues grow and dissipate, input–output
analysis updates these values during each time step.
• RO(i, t, p, k)—maximum flow (pc/ts) allowed to leave the ramp roadway
on segment i during time step t in analysis period p using branch k, due to
limited available storage at the downstream ramp terminal.
Methodology
The methodology for evaluating off-ramp queue spillback is integrated with
the freeway facility oversaturated segment procedure given in Chapter 25.
Exhibit 38-A8 depicts the methodology, highlighting additions and changes to
the Chapter 25 methodology to address off-ramp queue spillback.
Exhibit 38-A8
Freeway Facility
Oversaturated Analysis
Procedure, Adapted for
Off-Ramp Queue Spillback
Evaluation
The equivalent capacity SCEQ (in pc/h) of segment i with N lanes and NQ
blocked lanes is estimated as:
Equation 38-A12 𝑆𝐶𝐸𝑄(𝑖, 𝑁, 𝑁𝑄) = 𝑆𝐶(𝑖, 𝑁 − 𝑁𝑄) × 𝐶𝐴𝐹𝐵𝐿
Exhibit 38-A10 presents an example of a basic 4-lane directional segment
operating in Regime 4 (2 blocked lanes). The capacity of the unblocked lanes will
be equivalent to the capacity of a 2-lane basic segment with a capacity
adjustment factor CAFBL of 0.50 (4 directional lanes with 2 blocked lanes).
Exhibit 38-A10
Equivalent Segment Capacity
for Unblocked Lanes When
Lane Blockage Occurs
For the segment of Exhibit 38-A10, capacity under ideal conditions is:
• c = 2,400 pc/h (capacity per lane), or
• SC = 9,600 pc/h (segment capacity).
where
KBUB(i, j) = background density at the unblocked lanes in segment i, when
queue spillback occurs at the downstream segment;
ED(i) = expected demand at segment i, as defined in Chapter 25;
OFRPCT(i) = rate of off-ramp flow and mainline flow at segment i; and
KB[v, c] = density of a segment with demand flow rate v and capacity c, as
provided by Chapter 12 (basic segments), Chapter 13 (weaving
segments), or Chapter 14 (merge and diverge segments).
Shoulder length. The available shoulder length must be input by the analyst
for queue spillback analysis and is stored in the parameter SL(i) for
oversaturated calculations.
Deceleration lane length. The deceleration lane length is provided by the
analyst for the analysis of diverge segments. It is stored in the parameter LD(i)
for oversaturated calculations.
Spillback queue storage length. The maximum storage length for off-ramp
queues on segment i is computed as a function of the segment length L(i), the
deceleration lane length LD(i), and the number of queued lanes NQ(i). Exhibit 38-
A11 provides guidance on measuring each of the components required for
Regimes 3 and 4.
Exhibit 38-A11
Maximum Off-Ramp Queue
Storage Length at Diverge
Segments with Regime 3 or 4
Queue Spillback and
No Shoulder Available
Exhibit 38-A12
Maximum Off-Ramp Queue
Storage Length at Diverge
Segments with Regime 3 or 4
Queue Spillback and
Shoulder Available
The ramp structure must be modeled from the downstream end towards the
upstream end:
• For the location farthest downstream, provide one node for each lane
group or movement on the approach.
• At each subsequent upstream change in alignment, provide one node for
each ramp roadway lane connecting to a distinct lane group downstream.
The data structure used in the computations should be adjusted according to
this branch structure. Most parameters in the Chapter 25 oversaturated segment
evaluation methodology are computed as three-dimensional arrays (i, t, p),
where i is the segment’s index within the freeway facility and t refers to a specific
time step within a given analysis period p. In the case of two-lane ramps that
need to be evaluated independently, an extra dimension k is added to the ramp
parameter arrays to account for the specific lane under analysis. Lanes are
numbered right from left; therefore, k = 1 indicates the right ramp lane and k = 2
indicates the left ramp lane.
Example 1. In this example, shown in Exhibit 38-A13, only one lane connects
the freeway exit to the entry leg of the downstream roundabout. Therefore, only
one node is required at each location (i.e., a single branch structure, with k = 1 at
all nodes).
Exhibit 38-A13
Node Structure for Example 1
Exhibit 38-A14
Node Structure for Example 2
Exhibit 38-A15
Node Structure for Example 3
lane when spillback occurs. However, not all lane drop exits experience Regime 4
queue spillback. Regime 4 also occurs more frequently in locations with more
aggressive driver behavior. Local information and driver behavior should be
taken into consideration in determining the prevailing regime at a given site.
For operational analyses of existing locations, it is recommended that the
analyst provide the expected spillback regime based on observed field
conditions. For planning-level purposes where no field data are available, Exhibit
38-A16 provides the expected queue spillback regime as a function of the number
of exiting lanes and driver aggressiveness.
Exhibit 38-A17
Queue Influence Area with
Increased Turbulence
The QIA length is based on the time needed by arriving drivers to react to
partial lane blockage and to adjust their speeds and positions. Research (A-1) has
shown that traffic speeds upstream of the back of queue are negatively affected
at a headway distance of 10.95 s. Therefore, the influence area represents the
distance traversed by a vehicle during 10.95 s with a speed consistent with the
traffic stream.
The length is estimated as a function of the segment free-flow speed (FFS), as
shown in Exhibit 38-A18. The exact location of the QIA varies as a function of the
queue length. QIA lengths are shorter than the ramp influence distance of 1,500
ft. However, the two concepts are very different and are used differently in
analyzing ramp operations: the ramp influence area is used to analyze
undersaturated conditions, while the QIA is used to analyze oversaturated
conditions. Because drivers can only detect a downstream queue visually, they
have shorter reaction times compared to arriving at undersaturated off-ramps,
where signing and navigation information is provided in advance and allows
drivers to adjust their position earlier.
When Regimes 3 or 4 occur and lane blockage occurs on the mainline, the
QIA is added to the queue length to determine the extent of spillback effects. If
an upstream node is located within the combined length of the queue and QIA, a
capacity adjustment factor CAFUP (Equation 38-A39) must be applied to account
for the spillback effects.
Exhibit 38-A19 Ramp FFS, SFR (mi/h) Single-Lane Ramps Two-Lane Ramps
Capacity of Ramp Roadways >50 2,200 4,400
(pc/h) >40–50 2,100 4,200
>30–40 2,000 4,000
≥20–30 1,900 3,800
<20 1,800 3,600
Note: FFS = free-flow speed.
Exhibit 38-A20
Freeway Ramp Speed–Flow
Curves
The ramp density at capacity RKC is not necessarily the same as the 45
pc/mi/ln value used for freeway mainline lanes. This parameter is used to
evaluate the queue density at the ramp roadway during oversaturated conditions.
The ramp density at capacity is found by dividing the capacity by the speed.
Exhibit 38-A21 lists RKC values as a function of the ramp FFS.
Ramp Free-Flow Speed Ramp Capacity Ramp Density at Capacity Exhibit 38-A21
(mi/h) (pc/h/ln) (pc/mi/ln) Ramp Density at Capacity as a
Function of Ramp FFS
55 2,200 40.0
50 2,100 42.0
45 2,100 46.7
40 2,000 50.0
35 2,000 57.1
30 1,900 63.3
25 1,900 76.0
20 1,900 90.0
15 1,800 120.0
where
ISTG(i, k) = total available storage length from branch k at the intersection of
segment i (ft),
Nm = number of lanes serving movement m at the intersection,
Lm = storage length for movement m at the intersection (ft),
M = number of movements at the approach, and
Lh = average vehicle spacing in a stationary queue (ft/veh), from
Equation 31-155.
Exhibit 38-A23
Selection of a Cycle Reference
Point to Determine the Initial
Number of Vehicles Within the
Approach
Exhibit 38-A24
Example Signalized
Intersection Approach from an
Off-Ramp
Phase duration and effective green time. The duration of each phase at the signal
can be fixed (pre-timed control), or variable (semi-actuated or actuated control).
For the former case, phase duration is known. For the latter, an average phase
duration is estimated as described in Chapter 31, Section 2. The effective green
time g for each phase can then be computed according to Equation 19-3,
reproduced here as:
Equation 38-A19 𝑔 = 𝐷𝑝 − 𝑙1 − 𝑙2
where
g = effective green time (s),
Dp = phase duration (s),
l1 = start-up lost time = 2.0 (s),
l2 = clearance lost time = Y + Rc – e (s),
Y = yellow change interval (s),
Rc = red clearance interval (s), and
e = extension of effective green = 2.0 (s).
Converting approach capacity from analysis periods to time steps. The
standard signalized intersection analysis is performed in 15-min analysis
periods, while the queue spillback evaluation requires a 15-s time step approach
compatible with the freeway facilities oversaturated methodology. Therefore, an
adjustment is necessary to calculate each movement’s capacity in 15-s intervals.
The cycle length C can be divided into n time steps, with a duration of 15 s
each, as seen in Exhibit 38-A25. If an integer number of time steps is not
obtained, the difference is included in the first time-step of the next cycle. Next,
green times for each time step from 1 to n are computed. This procedure must be
repeated for every time step within the 15-min analysis period, resulting in a
total of 900 / 15 = 60 time steps.
Exhibit 38-A25
Assignment of Green Times to
Time Steps
The capacity ID for each movement, corresponding to a branch, for each time
step is obtained by multiplying the movement’s green time by its capacity:
𝐼𝐷(𝑖, 𝑡, 𝑝, 𝑘) = 𝑁𝑘 𝑠𝑘 𝐺𝑇(𝑖, 𝑡, 𝑝, 𝑘) × 𝑓𝐻𝑉 Equation 38-A20
where
ID(i, t, p, k) = discharge capacity from branch k in segment i during time step t
in analysis period p (veh/ts),
Nk = number of lanes serving movement k,
sk = saturation flow rate for movement k (veh/h/ln),
GT(i, t, p, l) = green time for lane group l from segment i during time step t in
analysis period p (s), and
fHV = adjustment factor for heavy vehicle presence (decimal).
The green time parameter GT(i, t, p, k) can range from 0 s (when the movement
has red through the entire time step length) to 15 s (when the movement has
green through the entire time step length). The heavy vehicle factor fHV is applied
to make the units used for intersection capacity (veh/h) consistent with the flow
rates used by the uninterrupted flow methods (pc/h).
Step 2H: Determine Reference Index for the Next Downstream Off-Ramp
This step is required to build this procedure’s computational engine, but it is
unimportant for understanding the overall methodology. The freeway facilities
methodology uses the parameter OFRF(i, t, p) to store the off-ramp flow rate at
diverge segment i. When a segment upstream of an off-ramp is evaluated for
queue spillback, the off-ramp flow rate must be referenced to estimate the
incoming flows for the blocked and non-blocked lanes. Therefore, a new variable
NEXTOFR(i) is introduced to reference the index of the closest diverge segment
downstream of segment i. This process is illustrated in Exhibit 38-A26, where the
node (i + 2) represents a diverge segment with an off-ramp flow vR. When the
queue extends upstream to node i, the approaching flow vf is split into two
groups: the exiting vehicles that will join the back of the queue, and the through
vehicles that will use the non-blocked lanes.
Exhibit 38-A26
Illustration of Mainline Flow
Rate Split into Blocked and
Unblocked Lanes
computed as the value from the previous time step, plus the difference between
demand RI and throughput RF at the ramp node. RUV is calculated as:
Equation 38-A25 𝑅𝑈𝑉(𝑖, 𝑡, 𝑝, 𝑘) = 𝑅𝑈𝑉(𝑖, 𝑡 − 1, 𝑝 , 𝑘) + 𝑅𝐼(𝑖, 𝑡, 𝑝, 𝑘) − 𝑅𝐹(𝑖, 𝑡, 𝑝, 𝑘)
where
RUV(i, t, p, k) = number of unserved passenger cars at the entrance of the ramp
roadway of segment i at the end of time step t during analysis
period p desiring to use branch k (pc);
RI(i, t, p, k) = maximum flow desiring to enter the off-ramp using branch k
of segment i during time step t in analysis period p, including
queues accumulated from previous time periods (pc/ts); and
RF(i, t, p, k) = flow that can enter the ramp roadway at segment i using
branch k during time step t in analysis period p (pc/ts).
If there are multiple branches k at the ramp roadway (e.g., two lane ramps),
RI and RF are compared for each branch k to obtain RUV for each branch k. The
total number of unserved passenger cars at the ramp RUV(i, t, p) is then obtained
as the sum of RUV for each branch:
𝐾
where
RO(i, t, p, k) = maximum flow allowed to leave the ramp roadway on
segment i during time step t in analysis period p using branch
k, due to limited available storage at the downstream ramp
terminal (veh/ts);
ISTG(i, k) = total available storage length from branch k at the intersection
of segment i (ft);
IUV(i, t, p, k) = number of unserved vehicles at the entrance of the intersection
of segment i using branch k at the end of time step t during
analysis period p (veh); and
ID(i, t, p, k) = discharge capacity (veh/ts) from branch k in segment i during
time step t in analysis period p (veh/ts).
Calculate intersection approach flow and number of unserved vehicles.
The intersection flow IF represents the number of vehicles that are able to cross
the boundary node between the ramp roadway and the intersection (i.e., its
capacity). It is computed as the smaller of the number of vehicles wishing to
enter the intersection and the maximum number of vehicles allowed to enter the
intersection due to the available queue storage at the intersection:
𝐼𝐹(𝑖, 𝑡, 𝑝, 𝑘) = min[𝐼𝐼(𝑖, 𝑡, 𝑝, 𝑘), 𝑅𝑂(𝑖, 𝑡, 𝑝, 𝑘)] Equation 38-A29
where IF(i, t, p, k) is the flow that can enter the intersection on segment i from
branch k during time step t in analysis period p (veh/ts), and other variables are
as defined previously.
If the number of vehicles trying to enter the intersection exceeds the amount
of vehicles allowed to enter the intersection, the number of total unserved
vehicles must be computed and considered in the intersection input II during the
next time period:
𝐼𝑈𝑉(𝑖, 𝑡, 𝑝, 𝑘) = 𝐼𝑈𝑉(𝑖, 𝑡 − 1, 𝑝, 𝑘) + 𝐼𝐼(𝑖, 𝑡, 𝑝, 𝑘) − 𝑅𝑂(𝑖, 𝑡, 𝑝, 𝑘) Equation 38-A30
𝑀𝑄2(𝑖, 𝑡, 𝑝, 𝑘) = 0
Exhibit 38-A27
Procedure for Evaluating the
Impact of Queue Spillback on
Upstream Nodes and
Determining the Queue
Length within Upstream
Segments
Exhibit 38-A28
Potential Effects of an Off-
Ramp Queue on Node i
with
𝑆𝐵𝐿𝐶(𝑖, 𝑡, 𝑝)
𝐿𝐶𝑅(𝑖, 𝑡, 𝑝) = Equation 38-A40
𝑆𝐹(𝑖, 𝑡, 𝑝)
where
CAFUP(i, t, p) = capacity adjustment factor for node i during time step t in
analysis period p;
LCR(i, t, p) = rate of lane change maneuvers in the QIA upstream of a queue
from an off-ramp, for segment i during time step t in analysis
period p;
SBLC(i, t, p) = number of lane change maneuvers within the QIA at node i
during time step t in analysis period p; and
SF(i, t, p) = segment flow out of segment i during time step t in analysis
period p (veh/ts).
The parameter LCR estimates the rate of lane change maneuvers performed
by drivers within the QIA trying to adjust their position when spillback occurs.
The parameter SBLC estimates the number of lane change maneuvers performed.
Exiting vehicles move to the shoulder lane to attempt to join the back of the queue,
while through vehicles move toward the median lanes to avoid the queue.
The computation of SBLC for a given node requires an estimate of the
number of vehicles in each freeway lane that plan to exit at the off-ramp. For
each lane 𝑖, the parameter pi represents the percentage of the off-ramp demand vR
traveling in the subject lane. The value of pi is a function of the distance from the
off-ramp to the subject node, as follows:
1. Within the ramp influence area (1,500 ft upstream from the diverge
point), the off-ramp demand flow rate vR is entirely positioned in the two
rightmost lanes, based on the Chapter 14 diverge segment methodology.
Therefore, the sum of the off-ramp flow rate percentages in the ramp
influence area p1,R and p2,R is equal to 1. The Appendix C methodology to
estimate lane-by-lane flow distribution in freeway segments is used to
estimate the lane flow ratio LFR for lanes 1 and 2. The values of p1,R and
p2,R are then estimated as follows:
𝐿𝐹𝑅1
𝑝1,𝑅 =
𝐿𝐹𝑅1 + 𝐿𝐹𝑅2
Equation 38-A41
𝐿𝐹𝑅2
𝑝2,𝑅 =
𝐿𝐹𝑅1 + 𝐿𝐹𝑅2
2. Beyond 8,000 ft upstream from the diverge point, the off-ramp has
negligible influence, again based on the Chapter 14 methodology.
Therefore, for any nodes located more than 8,000 ft from the off-ramp, pi
is assumed to be equally distributed among all N freeway lanes:
1
Equation 38-A42 𝑝𝑖 =
𝑁
3. Between 1,500 and 8,000 ft upstream from the diverge point, pi can be
obtained through linear interpolation between the first two cases as a
function of the distance from the diverge point dOFR in feet, as given by
Equation 38-A43. Exhibit 38-A29 illustrates the distribution of pi for a 3-
lane freeway segment.
Equation 38-A43 1
(𝑁 − 𝑝1,𝑅 ) × (𝑑𝑂𝐹𝑅 − 1,500)
𝑝𝑖 = 𝑝𝑖,𝑅 +
6500
Exhibit 38-A29
Distribution of pi as Function
of Distance from the Diverge
Point, for a 3-Lane Segment
Exhibit 38-A30
Illustration of Lane-Change
Maneuvers within the Queue
Influence Area in a 4-Lane
Segment under Regime 3
Exhibit 38-A31
Illustration of Lane-Change
Maneuvers within the Queue
Influence Area in a 4-Lane
Segment under Regime 4
where
ONRO(i, t, p) = maximum output flow rate that can enter the merge point
from on-ramp i during time step t in analysis period p;
RM(i, t, p) = maximum allowable rate of an on-ramp meter at the on-ramp
at node i during time step t in analysis period p;
ONRC(i, t, p) = geometric carrying capacity of the on-ramp at node i during
time step t in analysis period p;
MF(i, t, p) = actual mainline flow rate that can cross node i during time step
t in analysis period p;
ONRF(i, t, p) = actual ramp flow rate that can cross on-ramp node i during
time step t in time interval p;
MO3(i, t, p) = maximum allowable mainline flow rate across node i during
time step t in time interval p, limited by the presence of
queued vehicles at the upstream end of segment i while the
queue clears from the downstream end of segment i;
SC(i, t, p) = maximum flow rate that can pass through segment i at the end
of time step t in analysis period p based strictly on traffic and
geometric properties;
N(i, p) = number of lanes on segment i in analysis period p; and
all other variables are as defined previously.
Exhibit 38-A32
Effect of Queue Spillback on
the Discharge Capacity of an
Upstream On-Ramp
Exhibit 38-A33
Illustration of Different
Density Values within One
Diverge Segment
If there are no spillback effects, the segment operates with a uniform density.
In this case, the constraints for the unblocked and blocked portions (MO2UB and
MO2BL, respectively) are calculated proportionately to the number of unblocked
and blocked lanes:
(1 − 𝑁𝑄(𝑖))
Equation 38-A49 𝑀𝑂2𝑈𝐵(𝑖, 𝑡, 𝑝) = 𝑀𝑂2(𝑖, 𝑡, 𝑝) ×
𝑁(𝑖)
𝑁𝑄(𝑖)
Equation 38-A50 𝑀𝑂2𝐵𝐿(𝑖, 𝑡, 𝑝) = 𝑀𝑂2(𝑖, 𝑡, 𝑝) ×
𝑁(𝑖)
where
MO2UB(i, t, p) = maximum number of passenger cars that can enter the
unblocked portion of segment i during time step t and analysis
period p due to the presence of a queue in the downstream
ramp segment;
MO2(i, t, p) = maximum allowable mainline flow rate across node i during
time step t in time interval p, limited by available storage on
segment i due to a downstream queue;
NQ(i) = number of blocked lanes if the off-ramp queue backs up into
the freeway mainline in segment i;
N(i) = number of lanes in segment i; and
MO2BL(i, t, p) = maximum number of passenger cars that can enter the blocked
portion of segment i during time step t and analysis period p
due to the presence of a queue in the downstream ramp
segment.
If node i operates under increased turbulence (i.e., the node is in the QIA),
the unblocked portion of segment i will operate similar to a regular segment.
Therefore, the component MO2UB is equal to MO2 but proportional to the
number of lanes in the unblocked portion:
(1 − 𝑁𝑄(𝑖))
𝑀𝑂2𝑈𝐵(𝑖, 𝑡, 𝑝) = 𝑀𝑂2(𝑖, 𝑡, 𝑝) × Equation 38-A51
𝑁(𝑖)
where all variables are as defined previously.
For the blocked portion of segment i, MO2BL is calculated as equal to MO2
proportional to the number of lanes in the blocked portion plus an additional
number of vehicles due to the presence of a partial queue. This additional number
of vehicles is obtained by the bold terms in the following equation, which takes
into account the difference between the queue spillback density RKQ and the
segment queue density KQ, multiplied by the queue length:
𝑁𝑄(𝑖)
𝑀𝑂2𝐵𝐿(𝑖, 𝑡, 𝑝) = 𝑀𝑂2(𝑖, 𝑡, 𝑝) × Equation 38-A52
𝑁(𝑖)
+ 𝑺𝑩𝑳𝑸(𝒊, 𝒕 − 𝟏, 𝒑) × 𝑵𝑸(𝒊, 𝒕 − 𝟏, 𝒑)
× [𝑹𝑲𝑸(𝑶𝑭𝑹𝑵𝑬𝑿𝑻(𝒊), 𝒕 − 𝟏, 𝒑) − 𝑲𝑸(𝒊 − 𝟏, 𝒕 − 𝟏, 𝒑)]
where KQ(i, t, p) is the queue density (pc/mi/ln) of segment i during time step t in
analysis period p, and all other variables are as defined previously.
If node i experiences lane blockage, the values of queue density must be
computed for both the unblocked KQUB and blocked KQBL portions of segment
i. For the unblocked portion, KQUB is calculated similarly to Equation 25-10 in
Chapter 25, but the inputs for segment flow SF and segment capacity SC are
replaced by their equivalent parameters SFUB and SCEQ:
𝐾𝑄𝑈𝐵(𝑖, 𝑡, 𝑝) = 𝐾𝐽 − [(𝐾𝐽 − 𝐾𝐶) × 𝑆𝐹𝑈𝐵(𝑖, 𝑡 − 1, 𝑝)]/𝑆𝐶𝐸𝑄(𝑖, 𝑝) Equation 38-A53
where
KQUB(i, t, p) = queue density in the unblocked portion of segment i during
time step t in analysis period p (pc/mi/ln),
KJ = facilitywide jam density (pc/mi/ln),
KC = ideal density at capacity (pc/mi/ln),
SFUB(i, t, p) = segment flow out of the unblocked portion of segment i during
time step t in analysis period p (pc/ts), and
SCEQ(i, p) = equivalent capacity of the unblocked portion of segment i in
analysis period p (pc/ts).
The queue density for the blocked portion is equal to the ramp queue density:
𝐾𝑄𝐵𝐿(𝑖, 𝑡, 𝑝) = 𝑅𝐾𝑄(𝑂𝐹𝑅𝑁𝐸𝑋𝑇(𝑖), 𝑡 − 1, 𝑝) Equation 38-A54
where MF(i, t, p) is the actual mainline flow rate (pc/h) that can cross node i
during time step t in analysis period p, and other variables are as previously
defined.
Step 25: Update Number of Passenger Cars in the Blocked Portion of the
Segment
The number of passenger cars in the blocked portion NVBL during increased
turbulence is updated based on the number of vehicles in the previous time step
and considers the number of passenger cars that are able to leave the current and
upstream segments:
𝑁𝑉𝐵𝐿(𝑖, 𝑡, 𝑝) = 𝑁𝑉𝐵𝐿(𝑖, 𝑡 − 1, 𝑝) + 𝑀𝐹𝐵𝐿(𝑖 − 1, 𝑡, 𝑝) + 𝑂𝑁𝑅𝐹(𝑖 − 1, 𝑡, 𝑝) Equation 38-A60
− 𝑀𝐹𝐵𝐿(𝑖, 𝑡, 𝑝) − 𝑂𝐹𝑅𝐹(𝑖, 𝑡, 𝑝)
where NVBL(i, t, p) is the number of passenger cars present on the blocked
portion of segment i at the end of time step t during analysis period p, and all
other variables are as previously defined.
where
ΔK(i, p) = additional density in the queued mainline lines in segment i
during analysis period p (pc/mi/ln);
S = number of time steps in an analysis period (integer);
NQ(i) = number of blocked lanes in segment i (integer);
L(i) = length of segment i (mi);
ΔNV (i, t, p) = additional number of passenger cars in the congested portion
of a segment i due to an off-ramp queue during time step t in
analysis period p (pc);
OFRLQ(i, t, p) = queue length of off-ramp unserved vehicles for diverge
segment i during time step t in analysis period p (ft);
SBKQ(i, t, p) = spillback queue density for segment i during time step t in
analysis period p (pc/mi/ln), defined as equal to the ramp
roadway density RKQ(i, t, p) of the downstream off-ramp
segment experiencing queue spillback during the same time
step t in analysis period p;
KB(i, t, p) = background density (veh/h/ln), from Chapter 25; and
fHV(i, p) = adjustment factor for heavy vehicle presence for segment i
during analysis period p.
Similar to the mainline, the flow in the ramp roadway is also aggregated:
𝑆
𝑇
Equation 38-A65 𝑅𝐹(𝑖, 𝑝, 𝑘) = ∑ 𝑅𝐹(𝑖, 𝑡, 𝑝, 𝑘)
𝑆
𝑡=1
where RF(i, p, k) is the flow that can enter the ramp roadway at segment i using
branch k in analysis period p (pc/h), and all other variables are as previously
defined.
The aggregated density at the ramp is calculated as the average of the
number of vehicles inside the ramp during the analysis period:
𝑆
1
Equation 38-A66 𝑅𝐾(𝑖, 𝑝, 𝑘) = ∑ 𝑅𝑁𝑉(𝑖, 𝑡, 𝑝, 𝑘)
𝑆
𝑡=1
where RK(i, p, k) is the ramp roadway density for segment i using branch k in
analysis period p (pc/mi/ln), and all other variables are as previously defined.
Finally, the average speed SR(i, p, k) on branch k of the ramp roadway of
segment i during analysis period p (mi/h) is obtained by dividing the total ramp
flow using the branch in the analysis period by its average density:
𝑅𝐹(𝑖, 𝑝, 𝑘)
Equation 38-A67 𝑆𝑅(𝑖, 𝑝, 𝑘) =
𝑅𝐾(𝑖, 𝑝, 𝑘)
where other variables are as previously defined.
REFERENCES
A-1. Elefteriadou, L., M. Armstrong, Y. Zheng and G. Riente. Highway These references can be found
in the Technical Reference
Capacity Manual (HCM) Systems Analysis Methodology. Federal Highway Library in Volume 4.
Administration, Washington, D.C., 2016.
A-2. University of Florida Transportation Institute; Cambridge Systematics,
Inc.; and A. Skabardonis. NCHRP Web-Only Document 290: Highway
Capacity Manual Methodologies for Corridors Involving Freeways and Surface
Streets. Transportation Research Board, Washington, D.C., 2020.
Queue spillback into an urban street intersection may occur when the
freeway merge segment has insufficient capacity to process the ramp’s demand.
Spillback may also occur in cases of ramp metering. This appendix presents the
methodology for determining whether spillback will occur from an on-ramp into
the upstream intersection.
As shown in the framework in Exhibit 38-B1, the methodology considers
signalized intersections, two-way and all-way STOP-controlled intersections, and
roundabouts. The procedure first estimates the demand approaching the on-
ramp (determined based on the upstream intersection’s configuration), and then
estimates the on-ramp’s capacity. The Chapter 10 freeway facilities methodology
for oversaturated conditions can estimate the resulting queue length; however,
the user must input the on-ramp demand flow rate.
DEMAND ESTIMATION
The first step in the methodology calculates the entering demand flow rate vR
at the on-ramp as a function of the upstream intersection configuration and
operations. Under low-demand conditions, the on-ramp demand flow rate is
calculated as the sum of the demands on each of the intersection approaches that
discharge into the ramp. However, if any of these movements operates over
capacity, the total throughput to the ramp will be constrained by the capacity of
these oversaturated movements. Hence, this check ensures that the on-ramp
demand is not overestimated. The analysis approach for each of four intersection
types is presented next.
Exhibit 38-B1
Procedure for Detecting
Spillback Occurrence at an
On-Ramp
If all movements operate below capacity, the on-ramp demand is the sum of
the movement demands. However, if any movement that discharges into the on-
ramp operates over capacity, the total throughput to the on-ramp will be lower
than the sum of the corresponding intersection movements.
In the case of movements not controlled by the traffic signal discharging into
the on-ramp, those movements’ demands must also be compared to their
capacities. The potential capacity cp,i of an unsignalized movement can be
computed by aggregating its saturation flow rates during different phases of a
cycle.
If the unsignalized movement is free-flowing and there are no other
conflicting movements discharging to the on-ramp, its saturation flow rate sFF is
obtained from Equation 19-8, reproduced below as Equation 38-B2, applying the
applicable adjustment factors:
Equation 38-B2 𝑠𝐹𝐹 = 𝑠0 𝑓𝑤 𝑓𝐻𝑉𝑔 𝑓𝑔 𝑓𝑏𝑏 𝑓𝑎 𝑓𝐿𝑈 𝑓𝐿𝑡 𝑓𝑅𝑇 𝑓𝐿𝑝𝑏 𝑓𝑅𝑝𝑏 𝑓𝑤𝑧 𝑓𝑚𝑠 𝑓𝑠𝑝
where
sFF = saturation flow rate for the unsignalized movement during free-flow
(veh/h/ln),
s0 = base saturation flow rate (pc/h/ln), and
all other adjustment factors are as described with Equation 19-8.
If the unsignalized movement must yield to a conflicting movement
discharging to the on-ramp, the permitted saturation flow rate sp is calculated
based on Equation 31-100, reproduced below as Equation 38-B3:
𝜆0 𝑒 −𝜆0 𝑡𝑐𝑔/3,600
Equation 38-B3 𝑠𝑝 =
1 − 𝑒 −𝜆0 𝑡𝑓ℎ/3,600
where
sp = permitted saturation flow rate for unsignalized movement (veh/h/ln),
λ0 = throughput of the conflicting movement (veh/h/ln),
tcg = critical headway = 4.5 (s), and
tfh = follow-up headway = 2.5 (s).
The throughput of the conflicting movement λ0 is determined as a function of
the flow profile of the respective conflicting movement. The effective green g of
the conflicting movement is divided into a queue service time gs and a green
extension time ge, each with a specific flow profile:
• If the conflicting movement occurs during the queue service time gs, λ0 is
equal to the saturation flow rate s of the conflicting movement.
• If the conflicting movement occurs during the green extension time ge, λ0
is equal to the arrival flow rate during the green qg (Equation 19-32) of the
conflicting movement.
Exhibit 38-B2
Schematic of Movements
Turning to an On-Ramp from
a TWSC Intersection
where
λRT = departure rate from major street right turn into the on-ramp (veh/h),
vRT = demand flow rate for the major street right turn (veh/h), and
sRT = saturation flow rate for a right-turn movement (veh/h).
Exhibit 38-B3
Schematic of Movements
Turning to an On-Ramp from
an AWSC Intersection
Case D: Roundabouts
The roundabouts methodology is based on calculating the potential capacity
of each approach, based on three main variables: the critical headway, the
follow-up headway, and the circulating flow (Equation 22-21 through Equation
22-23). Critical and follow-up headway values can be obtained from Chapter 33,
Roundabouts: Supplemental. The methodology considers each approach
independently. To analyze roundabouts within a network, it is first necessary to
estimate the on-ramp throughput from a roundabout.
The procedure first identifies the movements that discharge to the on-ramp
and their respective ranks (priority orders). Exhibit 38-B4 illustrates a typical
roundabout, where movements discharging into the on-ramp are numbered
according to their ranks. In contrast to other types of intersections, the approach
furthest from the on-ramp has priority as it enters the circulating stream without
any significant conflicting traffic (other than occasional U-turns).
Exhibit 38-B4
Schematic of Movements
Turning to an On-Ramp from
a Roundabout
Rank 1 Movement (Left Turn from the Third Upstream Approach from the
On-Ramp)
This movement has priority over the other movements because it enters the
circulating stream first. In addition, because the on-ramp does not have an
approach into the roundabout, this movement is most often unopposed by the
circulating stream (except for occasional U-turns in the intersection). Therefore,
the maximum throughput 𝜆1−4 (veh/h) for this left-turn movement is given by:
Equation 38-B9 𝜆1−4 = min(𝑣1−4 , 𝑐1 )
where
λ1–4 = departure rate from the third upstream approach into the on-ramp
(veh/h),
v1–4 = demand flow rate for the third upstream approach into the on-ramp
(veh/h), and
c1 = potential capacity for the third upstream (rank 1) approach (veh/h).
Rank 2 Movement (Through from the Second Upstream Approach, Most Likely an
Off-Ramp):
This movement’s maximum throughput is limited by the upstream approach
departure rate and its own potential lane capacity c2, as defined in Equations 22-
21 through 22-23. Therefore, its maximum throughput is given by:
Equation 38-B10 𝜆2−4 = min(𝑣2−4 , 𝑐2 )
where
λ2–4 = departure rate from the second upstream approach into the on-ramp
(veh/h),
v2–4 = demand flow rate for the second upstream approach into the on-ramp
(veh/h), and
c2 = potential capacity for the second upstream (rank 2) approach (veh/h).
CAPACITY ESTIMATION
The on-ramp’s capacity is estimated in order to predict the occurrence of
queue spillback. The maximum output flow rate ONRO(i, t, p) that can enter the
merge point from on-ramp i during time step t in analysis period p can be
constrained by (a) the ramp metering rate, if ramp metering is active, or (b)
oversaturated conditions in the downstream merge segment.
Signalized Intersections
Exhibit 38-B5 presents the core methodology for evaluating the performance
of signalized intersections, modified to address the effects of on-ramp queue
spillback. The new and modified steps to the methodology are described below.
Exhibit 38-B5
Signalized Intersection
Methodology With
Adjustments to Address On-
Ramp Queue Spillback
Exhibit 38-B6
Typical Signalized Intersection
Ramp Terminal in a Diamond
Interchange
The total throughput from the intersection into the on-ramp λONR in veh/h is
the sum of the throughput from each of the contributing movements:
𝜆𝑂𝑁𝑅 = 𝜆𝑆𝐵𝐿 + 𝜆𝐸𝐵𝑇 + 𝜆𝑁𝐵𝑅 Equation 38-B13
The throughput for each movement i is the minimum value of its demand
and capacity:
𝜆𝑖 = min(𝑣𝑖 , 𝑐𝑖 ) Equation 38-B14
where
vi = demand flow rate for intersection movement i (veh/h), and
ci = capacity for intersection movement i (veh/h), from Equation 19-16.
Unsignalized movements, which are common for right-turn movements to
an on-ramp, are unrestricted. The capacity of these movements can be estimated
as the saturation flow rate (Equation 19-8) multiplied by the adjustment factor for
right turns fRT (Equation 19-13).
If all movements at the intersection are undersaturated, (i.e., vi ≤ ci for every
movement i), then Equation 38-B13 is simplified and the total on-ramp demand
throughput λONR is as follows:
Step 7B: Obtain Merging Capacity Using the Freeway Facilities Methodology
This step computes the merging capacity into the freeway cmerge. Three
potential bottlenecks can limit the on-ramp discharge into the freeway:
• The on-ramp capacity (Exhibit 14-12 or Exhibit 38-A2);
• The merge segment capacity, when the freeway facility is oversaturated; or
• The ramp metering rate, when ramp metering is active.
The procedure to obtain cmerge is presented in Exhibit 38-B7. The freeway
facility must be analyzed using the Chapter 10 methodology to evaluate whether
Exhibit 38-B7
Estimation of Freeway On-
Ramp Merging Capacity
Exhibit 38-B8
Sample Intersection for
Calculation of a QAP for the
On-Ramp
Exhibit 38-B9
On-Ramp Queue
Accumulation Polygon During
Queue Spillback
The cycle starts with the SBL green discharging into the on-ramp at a
throughput rate λSBL, while the on-ramp discharges to the freeway merge at a
rate cmerge. Therefore, the number of vehicles within the on-ramp grows at a rate
equal to (λSBL − cmerge). When the number of vehicles along the on-ramp reaches
the maximum ramp storage length LONR, vehicles from the intersection can only
be discharged to the on-ramp at the same the rate they are discharged from the
on-ramp into the freeway. The number of vehicles within the on-ramp is then
maintained and it is equal to LONR until the end of the green for the SBL
movement. At the end of the SBL green, the vertical difference between the
projected number of vehicles (dashed line) and the actual number of vehicles
inside the on-ramp represent the number of unserved vehicles for the SBL
approach. This additional queue can be considered in a multiperiod analysis for
the signalized intersection or interchange, using the methods provided in
Chapter 23, Ramp Terminals and Alternative Intersections.
The slope of the red line connecting the number of vehicles at the end and at
the start of green represent the reduced capacity of the SBL movement due to
queue spillback. For the remainder of the cycle, the NBR movement discharges at
a constant rate into the on-ramp, as this is an unsignalized movement. Given that
the discharge capacity cmerge is greater than the on-ramp demand λNBR, the vehicles
along the on-ramp are discharged to the freeway until the on-ramp is cleared.
Therefore, the NBR movement’s capacity is not affected by queue spillback.
This procedure can be applied for both pretimed and actuated control types,
because the core methodology can address both controller types. If the signal is
actuated, the average phase durations are applied, as obtained in Step 6.
Exhibit 38-B10
Illustration of Cooperative
Behavior in Unsignalized
Intersections with Queue
Spillback
Exhibit 38-B11
TWSC intersections Core
Methodology with
Adjustments to Address On-
Ramp Queue Spillback
i discharging into the on-ramp, the throughput is the minimum value of its
demand and the movement capacity:
𝜆𝑖 = min(𝑣𝑖 , 𝑐𝑚,𝑖 ) Equation 38-B20
where
vi = demand flow rate for movement i, and
cm,i = movement capacity for movement i (Equations 20-36, 20-37, and 20-40).
Step 9B: Obtain Merging Capacity Using the Freeway Facilities Methodology
This step computes the merging capacity into the freeway cmerge. The
procedure described in Step 7B of the queue spillback analysis for signalized
intersections (Exhibit 38-B5) is applied.
Exhibit 38-B12
On-Ramp Queue
Accumulation Polygon: TWSC
Intersection
From the relationship shown in Exhibit 38-B12, the spillback time Tsp is
defined as the amount of time within an analysis period when spillback occurs:
𝐿𝑂𝑁𝑅 − 𝑁(0)
Equation 38-B21 𝑇𝑠𝑝 = 𝑇 −
𝜆𝑂𝑁𝑅 − 𝑐merge
where
Tsp = amount of time with active spillback (min),
T = analysis period duration (min),
LONR = available queue storage at the on-ramp (veh),
N(0) = number of queued vehicles along the on-ramp at t = 0 (start of the
cycle),
λONR = discharge from the intersection into the on-ramp (veh/h), and
cmerge = merging capacity of the on-ramp (veh/h).
Estimating the spillback time Tsp is a key element of the methodology,
because the aggregated calculations of capacity for each movement depend on
the amount of time that the intersection operates under queue spillback.
3,600 𝜆𝑂𝑁𝑅
𝑐𝐸𝑄,𝑖 × 𝑐𝐸𝑄,𝑖
2
3,600 𝑣𝑖 √ 𝑣𝑖
𝑑𝑠𝑝 = + 900𝑇 −1+ ( − 1) + +5 Equation 38-B25
𝑐𝐸𝑄,𝑖 𝑐𝐸𝑄,𝑖 𝑐𝐸𝑄,𝑖 450𝑇
[ ]
where all variables are as previously defined.
Exhibit 38-B13
AWSC Intersection Core
Methodology with
Adjustments to Address
On-Ramp Queue Spillback
Roundabouts
The methodology presented in Chapter 22, Roundabouts, is shown in Exhibit
38-B14. The steps added to the methodology to evaluate queue spillback effects
are shown in red. Each of the new steps is discussed in the subsections below.
This methodology is applicable only to single-lane roundabouts. Exhibit 22-9
provides the required input data and potential data sources for the core
roundabout methodology. Exhibit 38-B15 lists the additional input data required
for queue spillback analysis.
Exhibit 38-B14
Roundabouts Methodology
With Adjustments to Address
On-Ramp Queue Spillback
Required Data and Units Potential Data Source(s) Suggested Default Exhibit 38-B15
On-Ramp Data Required Data and Potential
Data Sources for Roundabout
On-ramp metering rate (veh/h) Design plans, field data Must be provided
Queue Spillback Evaluation
On-ramp storage length LONR (ft) Field data Must be provided
Roundabout Data
Departure saturation headway into
Field data 3 s/veh
the on-ramp hs (s/veh)
Step 13: Compute the Maximum Throughput Into the On-Ramp for Each O-D
Movement
The maximum throughput into the on-ramp for each movement is calculated
using the roundabout priority order, starting with the most upstream approach
from the on-ramp exit leg and proceeding counterclockwise. The Rank 1
approach (Exhibit 38-B16) is the one whose flow has the highest priority, given
that it enters the circulating stream upstream of all other approaches. The next-
highest priority movement is the Rank 2 approach, and the lowest-priority
movement is the Rank 3 approach.
Exhibit 38-B16
Example Priority Order for a
Roundabout Upstream of an
On-Ramp
Next, the methodology calculates the capacity of the roundabout’s exit lane
into the on-ramp. Research (B-2, B-3) suggests that the capacity of an exit lane,
accounting for pedestrian and bicycle traffic in a typical urban area, is in the
range of 1,200 to 1,300 veh/h. Starting from the Rank 1 approach, and proceeding
counterclockwise with the remaining approaches, each approach’s capacity is
used to determine the maximum throughput for every movement discharging to
the on-ramp.
Rank 1 Approach
The Rank 1 approach (the SB approach in the example in Exhibit 38-B16) has
priority over the other movements connecting to the on-ramp because it enters
the circulating stream first. In addition, because the on-ramp leg usually does not
have an approach into the roundabout, the Rank 1 movement is most often
unopposed by the circulating stream (except for occasional U-turns along the
arterial). Therefore, the maximum throughput λSB-ONR for this left-turn movement
is limited by its own lane capacity cSB and the maximum throughput to the on-
ramp:
3,600
Equation 38-B27 𝜆𝑆𝐵−𝑂𝑁𝑅 = min (𝑣𝑆𝐵−𝑂𝑁𝑅 , 𝑐𝑆𝐵 × 𝑝𝑆𝐵−𝑂𝑁𝑅 , )
ℎ𝑠
with
𝑣𝑆𝐵−𝑂𝑁𝑅
Equation 38-B28 𝑝𝑆𝐵−𝑂𝑁𝑅 =
𝑣𝑆𝐵𝑈 + 𝑣𝑆𝐵−𝑂𝑁𝑅 + 𝑣𝑆𝐵𝑇
where
λSB-ONR = departure rate from the SB approach into the on-ramp (veh/h);
vSB-ONR = demand flow rate for the SB approach into the on-ramp (veh/h);
cSB = lane capacity for the SB approach, from Equation 22-21 (veh/h);
Rank 2 Approach
The maximum throughput for the Rank 2 movement (the EB approach in the
example in Exhibit 38-B16) is limited by its own lane capacity cEB, as defined in
Equations 22-21 through Equation 22-23, and the maximum throughput after
considering the departure rate of the upstream Leg 1. Therefore, the maximum
throughput λEB-ONR for this movement is given by:
3,600
𝜆𝐸𝐵−𝑂𝑁𝑅 = min (𝑣𝐸𝐵−𝑂𝑁𝑅 , 𝑐𝐸𝐵 × 𝑝𝐸𝐵−𝑂𝑁𝑅 , − 𝜆𝑆𝐵−𝑂𝑁𝑅 ) Equation 38-B29
ℎ𝑠
with
𝑣𝐸𝐵−𝑂𝑁𝑅
𝑝𝐸𝐵−𝑂𝑁𝑅 = Equation 38-B30
𝑣𝐸𝐵𝐿 + 𝑣𝐸𝐵−𝑂𝑁𝑅 + 𝑣𝐸𝐵𝑅
where
λEB-ONR = departure rate from the EB approach into the on-ramp (veh/h);
vEB-ONR = demand flow rate for the EB approach into the on-ramp (veh/h);
cEB = lane capacity for the EB approach, from Equation 22-21 (veh/h);
Rank 3 Approach
Similar to Rank 2 movements, the maximum throughput for the Rank 3
movement (the NB approach in the example in Exhibit 38-B16) is limited by its
own lane capacity cNB, as defined in Equation 22-21 through Equation 22-23, and
the maximum throughput to the on-ramp after considering departure rates from
the upstream approaches. Therefore, the maximum throughput λNB-ONR for this
right-turn movement is given by:
3,600
𝜆𝑁𝐵−𝑂𝑁𝑅 = min (𝑣𝑁𝐵−𝑂𝑁𝑅 , 𝑐𝑁𝐵 × 𝑝𝑁𝐵−𝑂𝑁𝑅 , − 𝜆𝑆𝐵−𝑂𝑁𝑅 − 𝜆𝐸𝐵−𝑂𝑁𝑅 ) Equation 38-B31
ℎ𝑠
with
𝑣𝑁𝐵−𝑂𝑁𝑅
𝑝𝑁𝐵−𝑂𝑁𝑅 = Equation 38-B32
𝑣𝑁𝐵𝑈 + 𝑣𝑁𝐵𝑇 + 𝑣𝑁𝐵−𝑂𝑁𝑅
where
λNB-ONR = departure rate from the NB approach into the on-ramp (veh/h);
vNB-ONR = demand flow rate for the NB approach into the on-ramp (veh/h);
cNB = lane capacity for the NB approach, from Equation 22-21 (veh/h);
Step 15: Compute On-Ramp Merging Capacity and Compare to the Maximum
Throughput to the On-Ramp
The on-ramp merging capacity is calculated exactly the same as in Step 7B of
the queue spillback methodology for signalized intersections (Exhibit 38-B5). The
maximum number of vehicles that can merge into the on-ramp cmerge (from
Equation 25-18) is compared to the maximum throughput from the roundabout
to the on-ramp λONR. If cmerge > λONR, then spillback is not expected to occur, and
no adjustments are necessary. If cmerge ≤ λONR, queues will develop along the on-
ramp, and spillback may occur if the queue storage is insufficient. In this case,
the analyst proceeds to Step 16 to evaluate the on-ramp queue storage ratio to
evaluate whether spillback will occur.
Step 16: Determine the On-Ramp Storage Ratio and Queue Spillback Length
Given the throughput from the roundabout into the on-ramp λONR, the queue
length QONR (in veh) along the on-ramp during a 15-min analysis period is:
𝜆𝑂𝑁𝑅 − 𝑐merge
Equation 38-B34 𝑄𝑂𝑁𝑅 =
4
where all variables are as previously defined.
If a multi-period analysis is performed, the queue length for the current analysis
period p must be added to the queue length from the previous analysis period:
𝜆𝑂𝑁𝑅,𝑝 − 𝑐merge,𝑝
Equation 38-B35 𝑄𝑂𝑁𝑅,𝑝 = 𝑄𝑂𝑁𝑅,𝑝−1 +
4
where all variables are as previously defined.
The on-ramp storage ratio is calculated by dividing the available on-ramp
storage length by the average vehicle spacing:
𝐿ℎ × 𝑄𝑂𝑁𝑅
Equation 38-B36 𝑅𝑄 =
𝐿𝑂𝑁𝑅
where
RQ = on-ramp storage ratio (decimal);
Lh = average vehicle spacing (ft/veh), from Equation 31-155;
QONR = on-ramp queue length (veh); and
LONR = on-ramp length (ft).
If the on-ramp storage ratio RQ is greater than 1, queues will form along each
roundabout approach due to spillback. The value of RQ is specific to a given
analysis period. If congestion is expected, but RQ < 1 for a single analysis period,
multi-period analysis may have to be conducted.
Step 18: Calculate the Average Delay by Approach and Aggregate to the Average
Control Delay
To estimate the average delay per approach, the delay due to the on-ramp
capacity limitation is estimated and added to the approach control delay
calculated by Equation 22-17. This equation assumes no residual queue at the
start of the analysis period. If queue spillback occurs, the average control delay is
significantly affected by the analysis period length. However, Chapter 22 does
not provide a multiperiod analysis method. Therefore, the delay results may not
be accurate when a queue exists at the start of the analysis period.
As an alternative, an iterative process that carries over queues from one time
period to the next (B-4) may be used. The additional delay (in sec/veh) due to the
on-ramp spillback is calculated as follows:
3,600 𝜆𝑂𝑁𝑅
𝑐merge × 𝑐merge
2
3,600 𝜆𝑂𝑁𝑅 𝜆
√ 𝑂𝑁𝑅
𝑑𝑠𝑝 = + 900𝑇 −1+ ( − 1) + Equation 38-B41
𝑐merge 𝑐merge 𝑐merge 450𝑇
[ ]
𝜆𝑂𝑁𝑅
+ 5 × min [ , 1]
𝑐merge
where all variables are as previously defined.
REFERENCES
Some of these references can B-1. Aakre, E., and A. Aakre. Modeling cooperation in unsignalized
be found in the Technical
Reference Library in Volume 4. intersections. Procedia Computer Science, Vol. 109, 2017, pp. 875–880.
B-2. Robinson, B., L. Rodegerdts, W. Scarbrough, W. Kittelson, R. Troutbeck,
W. Brilon, L. Bondzio, K. Courage, M. Kyte, J. Mason, A. Flannery, E.
Myers, and J. Bunker. Roundabouts: An Informational Guide. Federal
Highway Administration, Washington, D.C., 2000.
B-3. Rodegerdts, L. A., and G. E. Blackwelder. Analytical Analysis of
Pedestrian Effects on Roundabout Exit Capacity. In Transportation Research
Circular E-C083, Transportation Research Board, Washington, D.C., 2005.
B-4. Kimber, R. M. and E. M. Hollis. Traffic queues and delays at road junctions.
Transport and Road Research Laboratory, Wokingham, Berkshire, U.K.,
1979.
Basic Segments
The calibration parameters a and b used to analyze basic segments are
computed as follows:
Equation 38-C3 𝑎 = 𝑎0 + 𝐺 × 𝑎𝐺 + 𝐻𝑉% × 𝑎𝐻𝑉 + 𝑟𝑑 × 𝑎𝑟𝑑
Equation 38-C4 𝑏 = 𝑏0 + 𝐺 × 𝑏𝐺 + 𝐻𝑉% × 𝑏𝐻𝑉 + 𝑟𝑑 × 𝑏𝑟𝑑
where
a = multiplicative calibration parameter;
a0 = empirical constant, from Exhibit 38-C1;
G = grade (%);
ag = empirical coefficient due to the effect of grade, from Exhibit 38-C1;
HV% = heavy vehicle percentage (%);
aHV = empirical coefficient due to the effect of trucks, from Exhibit 38-C1;
rd = ramp density = total number of ramps 0.5 mi upstream and 0.5 mi
downstream of the segment;
ard = empirical coefficient due to the effect of ramp density, from Exhibit
38-C1;
b = additive calibration parameter;
b0 = empirical constant, from Exhibit 38-C1;
bg = empirical coefficient due to the effect of grade, from Exhibit 38-C1;
bHV = empirical coefficient due to the effect of trucks, from Exhibit 38-C1; and
brd = empirical coefficient due to the effect of ramp density, from Exhibit
38-C1.
Exhibit 38-C1 provides the coefficients used in the LFD model for basic,
merge, and diverge segments. The coefficients are specific for each combination
of lane number, segment type, and total number of lanes in the segment.
Lane Para- Basic Segments Diverge Segments Merge Segments Exhibit 38-C1
# meter 2 lanes 3 lanes 4 lanes 2 lanes 3 lanes 4 lanes 2 lanes 3 lanes 4 lanes Lane Flow Distribution Model
a0 0.18 0.027 0.068 0.0097 −0.075 0.31 0.015 0.0029 −0.077 Coefficients for Basic, Merge,
b0 and Diverge Segments
0.52 0.27 0.22 0.44 0.27 0.25 0.59 0.28 0.24
ag 0.024 0.021 −0.011 0.0097 0.0077 −0.034 0.015 −0.0029 −0.0030
aHV −0.048 −0.0036 −0.0021 −0.0093 0.00080 −0.057 −0.0093 −0.0029 0.011
ard −0.095 −0.0083 −0.059 −0.0097 0.014 −0.028 −0.0047 −0.0029 0.014
L1
bg 0.0030 0.0097 −0.034 −0.0098 −0.0081 −0.00016 0.020 0.031 0.040
bHV 0.008 −0.0029 0.0024 0.0078 0.0014 −0.019 −0.014 −0.0018 −0.027
brd 0.0013 0.032 −0.035 0.00057 0.031 0.0052 −0.040 −0.042 −0.041
avr −0.21 −0.067 −0.0087 −0.035 −0.10 0.026
bvr −0.13 0.013 −0.021 −0.070 −0.030 0.0091
a0 −0.063 −0.025 0.0096 0.29 −0.0082 −0.080
b0 0.31 0.29 0.34 0.25 0.38 0.24
ag −0.0060 0.0015 −0.0096 −0.035 −0.0082 0.00048
aHV 0.0011 0.00027 −0.00054 −0.052 −0.00082 0.013
ard 0.0037 −0.0085 −0.0096 −0.030 −0.0026 0.018
L2
bg −0.017 −0.024 −0.0019 0.0019 0.0079 −0.019
bHV 0.0024 −0.00036 0.00089 −0.0041 −0.00048 −0.0067
brd 0.01 −0.041 0.0052 0.0044 −0.0060 0.0010
avr −0.048 −0.0065 −0.12 −0.033
bvr −0.073 −0.0091 −0.039 −0.013
a0 −0.045 0.27 0.029
b0 0.28 0.25 0.25
ag −0.0017 −0.036 −0.0017
aHV 0.0021 −0.044 −0.0058
ard 0.0081 −0.034 −0.0068
L3
bg 0.011 0.0034 0.00060
bHV −0.0011 0.0092 0.014
brd 0.015 0.0016 0.018
avr 0.021 −0.079
bvr −0.0064 −0.041
Note: Empty cells indicate the factor is not used in the model for the given combination of lane number, segment
type, and number of lanes in the segment (i.e., substitute a value of 0 for the coefficient).
Weaving Segments
The calibration parameters a and b used to analyze weaving segments are
computed by Equation 38-C7 and Equation 38-C8. The coefficients in these
equations include factors addressing the effects of weaving-specific properties.
Equation 38-C7
𝑣𝑅,𝑚 𝑣𝑅,𝑑
𝑎 = 𝑎0 + 𝐺 × 𝑎𝐺 + 𝐻𝑉% × 𝑎𝐻𝑉 + 𝐼𝐷 × 𝑎𝐼 + × 𝑎𝑣𝑚 + × 𝑎𝑣𝑑
1,000 1,000
𝐿𝑠
+ × 𝑎𝐿𝑆 + 𝑉𝑅 × 𝑎𝑉𝑅
1,000
𝑣𝑅,𝑚 𝑣𝑅,𝑑
Equation 38-C8 𝑏 = 𝑏0 + 𝐺 × 𝑏𝐺 + 𝐻𝑉% × 𝑏𝐻𝑉 + 𝐼𝐷 × 𝑏𝐼 + × 𝑏𝑣𝑚 + × 𝑏𝑣𝑑
1,000 1,000
𝐿𝑠
+ × 𝑏𝐿𝑆 + 𝑉𝑅 × 𝑏𝑉𝑅
1,000
where
Chapter 13, Freeway Weaving
Segments, defines interchange ID = interchange density (interchanges/mi);
density as the number of
interchanges within 3 mi aI = empirical coefficient due to effect of interchange density, from Exhibit
upstream and downstream of 38-C2;
the center of the subject
weaving segment, divided vR.m = on-ramp flow (veh/h);
by 6.
avm = empirical coefficient for on-ramp flow, from Exhibit 38-C2;
vR.d = off-ramp flow (veh/h);
avd = empirical coefficient for off-ramp flow, from Exhibit 38-C2;
Ls = weaving segment length (ft);
aLS = empirical coefficient for weaving segment length, from Exhibit 38-C2;
VR = volume ratio = weaving volume / total volume (decimal);
aVR = empirical coefficient for volume ratio, from Exhibit 38-C2;
bI = empirical coefficient due to effect of interchange density, from Exhibit
38-C2;
bvm = empirical coefficient for on-ramp flow, from Exhibit 38-C2;
bvd = empirical coefficient for off-ramp flow, from Exhibit 38-C2;
bLS = empirical coefficient for weaving segment length, from Exhibit 38-C2;
bVR = empirical coefficient for volume ratio, from Exhibit 38-C2; and
all other variables are as previously defined.
Exhibit 38-C2 provides the coefficients used in the LFD model for weaving
segments. The coefficients are specific for each combination of lane number,
segment type, and total number of lanes in the segment.
Exhibit 38-C3
LFR Distribution for a Sample
2-Lane Basic Freeway
Segment
Exhibit 38-C4 illustrates the LFR distribution for a 3-lane segment. At low
demand, most of the flow is concentrated in the center lane (Lane 2), followed by
Lane 1 and Lane 3. As demand increases, the LFR increases in Lane 3 and
decreases in Lanes 1 and 2.
Exhibit 38-C4
LFR Distribution for a Sample
3-Lane Basic Freeway
Segment
Exhibit 38-C5 shows the LFR distribution for a 4-lane segment. Under free-
flow conditions, Lanes 2 and 3 carry the majority of flow. Lane 4 is typically
underused during undersaturated conditions, but at higher demands it carries
the majority of the flow.
Exhibit 38-C5
LFR Distribution for a Sample
4-Lane Basic Freeway
Segment
The flow distribution patterns shown in the exhibits above for basic
segments are also observed in merge, diverge, and weaving segments.
Additional factors such as ramp volume, grade, and truck percentage influence
the boundary values and slopes of the curves, but do not change the typical LFR
distribution as a function of the v/c ratio.
Exhibit 38-C6
Check for Negative Lane
Flows
Exhibit 38-C7
Check for Lane Capacity
Lane-by-Lane FFS
Field observations have shown that speeds differ among lanes, with speeds
typically lower in shoulder lanes and higher in median lanes. The model for
estimating individual lane FFS applies a multiplicative factor xFFS to the segment
FFS as follows:
Equation 38-C9 𝐹𝐹𝑆𝑖 = 𝐹𝐹𝑆𝑎𝑑𝑗 × 𝑥𝐹𝐹𝑆
where
FFSi = free-flow speed for lane i (mi/h);
FFSadj = adjusted free-flow speed for the segment (mi/h), from Equation 12-5;
and
xFFS = FFS multiplier, from Exhibit 38-C8.
Exhibit 38-C8 presents the recommended multipliers, which are provided as
a function of the segment type and the number of lanes in the segment. As shown,
when the number of lanes increases, the range of FFS multipliers increases as
well (i.e., lower speeds exist in the shoulder lanes and higher speeds exist in the
median lanes). For 2-lane segments, merge and diverge segments have a higher
FFS differential between the two lanes, compared to basic segments. For 3-lane
segments, basic segments have the greatest FFS range, while merge segments
have more uniform lane FFS. Finally, for 4-lane segments, merge segments have
the greatest FFS range, while basic and merge segments have similar ranges.
Lane-by-Lane Capacity
Basic, Merge, and Diverge Segments
Similar to the case of free-flow speeds, capacities also differ among lanes, with
capacities typically lower in shoulder lanes and higher in median lanes. Center
lanes typically have values similar to the segment average. The model for
estimating individual lane FFS in basic, merge, and diverge segments applies a
multiplicative factor xc to the segment capacity as follows:
𝑐𝑖 = 𝑐𝑎𝑑𝑗 × 𝑁 × 𝑥𝑐 Equation 38-C10
where
ci = capacity of lane i (pc/h);
cadj = adjusted capacity for the segment (pc/h/ln), from Equation 12-8;
N = number of lanes in the segment; and
xc = capacity multiplier, from Exhibit 38-C9.
Exhibit 38-C9 presents the percent of the total segment capacity distributed
to each lane in the segment, defining a capacity multiplier xc for each
combination of segment type and number of lanes.
Segment capacities measured from field data may not equal the estimated
capacities from the Chapter 12 methodology for basic freeway segments. Field
measurements of capacity have been found to be lower than HCM estimates (C-3).
Weaving Segments
Capacity distributions are observed to be significantly more complex in
weaving segments compared to other types of freeway segments, and the
breakdown method does not provide reliable results. Capacity is assumed to be
uniform for all lanes within a weaving segment, and is obtained from Equation
13-5, adapted here as Equation 38-C11, based on a maximum density of 43 pc/h/ln:
Equation 38-C11 𝑐𝐼𝑊𝐿 = 𝑐𝐼𝐹𝐿 − [438.2 (1 + 𝑉𝑅)1.6] + (0.0765 𝐿𝑆 ) + (119.8𝑁𝑊𝐿 )
where
cIWL = per-lane capacity of the weaving segment under equivalent ideal
conditions (pc/h/ln),
cIFL = per-lane capacity of a basic freeway segment with the same FFS as the
weaving segment under equivalent ideal conditions (pc/h/ln),
VR = volume ratio,
LS = weaving segment length (ft), and
NWL = number of lanes from which weaving maneuvers may be made with
either one or no lane changes.
Lane-by-Lane Speed
With flow, capacity, and FFS determined by lane, the speed–flow model for
freeway segments given by Equation 12-1 and Exhibit 12-6 is then adapted to
estimate the speeds in individual lanes. Speed in each lane i is determined as:
𝑆𝑖 = 𝐹𝐹𝑆𝑖 if 𝑣𝑖 ≤ 𝐵𝑃𝑖
Equation 38-C12
𝑐
(𝐹𝐹𝑆𝑖 − 𝑖 ) (𝑣𝑖 − 𝐵𝑃𝑖 )2
𝑆𝑖 = 𝐹𝐹𝑆𝑖 − 45 if 𝑣𝑖 > 𝐵𝑃𝑖
(𝑐𝑖 − 𝐵𝑃𝑖 )2
with
Equation 38-C13 𝑣𝑖 = 𝑣 × 𝐿𝐹𝑅𝑖
where
Si = speed in lane i (mi/h);
FFSi = free-flow speed for lane i (mi/h);
ci = capacity of lane i (pc/h/ln);
vi = demand flow rate for lane i (pc/h/ln);
BPi = breakpoint value for lane i (pc/h/ln), from Equation 38-C14;
v = demand flow rate for the segment (pc/h); and
LFRi = share of the total flow on lane i.
where
BPi = breakpoint value for lane i (pc/h/ln),
FFSi = free-flow speed for lane i (mi/h), and
CAF = capacity adjustment factor, from Exhibit 12-6.
For auxiliary lanes in weaving segments, individual lane speeds cannot be
addressed by this methodology because conditions vary widely along the
auxiliary lane’s length, as discussed previously. For the O-D analysis described
in Section 3 of this chapter, auxiliary lane speeds are only relevant when the
subject weaving segment is the entry or exit point of a freeway facility for a
particular O-D pair. In this case, the expected speed for the segment described by
Equation 38-9 can be replaced by the average speed of weaving vehicles
described in Chapter 13:
𝐹𝐹𝑆 × 𝑆𝐴𝐹 − 15
𝑆𝑊 = 15 + ( ) Equation 38-C15
1+𝑊
with
𝐿𝐶ALL 0.789
𝑊 = 0.226 + ( ) Equation 38-C16
𝐿𝑆
where
SW = average speed of weaving vehicles within the weaving segment (mi/h),
FFS = free-flow speed of the weaving segment (mi/h),
SAF = speed adjustment factor,
W = weaving intensity factor (unitless),
LCALL = speed adjustment factor (unitless), and
LS = weaving intensity factor (unitless).
APPLICATION EXAMPLES
5,003
𝐿𝐹𝑅2 = −0.0568 × ln ( ) + 0.275 = 0.296
7,200
Finally, the LFR for the leftmost lane (lane 3) is obtained from Equation 38-C2:
𝐿𝐹𝑅3 = 1 − 𝐿𝐹𝑅2 − 𝐿𝐹𝑅1 = 1 − 0.350 − 0.296 = 0.354
Exhibit 38-C10
Comparison of Speed–Flow
Curves by Lane and for the
Segment
Exhibit 38-C11
Example of LFR Calculation for
a Weaving Segment
The heavy vehicle adjustment factor is calculated from Equation 12-10, using
a passenger car equivalency ET of 2:
1 1
𝑓𝐻𝑉 = = = 0.968
1 + 𝑃𝑇 (𝐸𝑇 − 1) 1 + 0.03(2 − 1)
The weaving and non-weaving demands are adjusted to flow rates under
ideal conditions. Because the demands are estimated based on 15-min intervals,
it is assumed the PHF is set to 1.
𝑉
𝑣=
𝑃𝐻𝐹 × 𝑓𝐻𝑉
24
𝑣𝑅𝑅 = = 24.8 pc/h
1 × 0.968
404
𝑣𝑅𝐹 = = 417.3 pc/h
1 × 0.968
1200
𝑣𝐹𝑅 = = 1,239.6 pc/h
1 × 0.968
3312
𝑣𝐹𝐹 = = 3,421.3 pc/h
1 × 0.968
The weaving and non-weaving flows are given by:
𝑣𝑊 = 𝑣𝐹𝑅 + 𝑣𝑅𝐹 = 1,239.6 + 417.3 = 1,656.9 pc/h
𝑣𝑁𝑊 = 𝑣𝑅𝑅 + 𝑣𝐹𝐹 = 24.8 + 3,421.3 = 3,446.1 pc/h
The volume ratio is computed as:
𝑣𝑊 1,656.9
𝑉𝑅 = = = 0.325
𝑣 1,656.9 + 3,446.1
A weaving segment’s capacity is the smaller of the density-based capacity
cIWL from Equation 13-5 and the weaving-demand-based capacity cIW from
Equation 13-7. The segment’s base capacity cIFL is 2,400 pc/h based on the
measured FFS of 70 mi/h. Because the LFR is calculated in the next step using
flows and capacities in veh/h/ln, the capacities calculated in this step are
converted from units of pc/h/ln by applying the heavy vehicle factor.
𝑐′𝐼𝑊𝐿 = 𝑐𝐼𝐹𝐿 − [438.2(1 + 𝑉𝑅)1.6 ] + (0.0765𝐿𝑠 ) + (119.8𝑁𝑊𝐿 )
2,400 2,400
𝑐′𝐼𝑊 = = = 7,385 pc/h
𝑉𝑅 0.325
𝑐′𝐼𝑊𝐿 × 𝑓𝐻𝑉 7,385 × 0.968
𝑐𝐼𝑊 = = = 1,787 veh/h/ln
𝑁UP 4
The capacity of the weaving segment upstream the weave is obtained by:
𝑐UP = 𝑐 × 𝑁UP = 1,787 × 4 = 7,148 veh/h
The flow ratio for lane 1 (right lane) is obtained from Equation 38-C1:
𝑣up
𝐿𝐹𝑅1 = 𝑎1 × ln ( ) + 𝑏1
𝑐UP
Because this a weaving segment, the calibration parameters 𝑎 and 𝑏 for lane 1
are obtained using Equation 38-C7, Equation 38-C8, and Exhibit 38-C2 as follows:
𝑣𝑅,𝑚 𝑣𝑅,𝑑
𝑎1 = 𝑎0 + 𝐺 × 𝑎𝐺 + 𝐻𝑉% × 𝑎𝐻𝑉 + 𝐼𝐷 × 𝑎𝐼 + × 𝑎𝑣𝑚 + × 𝑎𝑣𝑑
1,000 1,000
𝐿𝑠
+ × 𝑎𝐿𝑆 + 𝑉𝑅 × 𝑎𝑉𝑅
1,000
𝑏1 = 0.178
The lane flow ratio for lane 1 is then:
4,512
𝐿𝐹𝑅1 = −0.181 × ln ( ) + 0.178
7,148
𝐿𝐹𝑅1 = 0.261
The same procedure is applied to estimate the LFR for lane 2, using the
respective coefficients from Exhibit 38-C2:
𝑣𝑅,𝑚 𝑣𝑅,𝑑
𝑎2 = 𝑎0 + 𝐺 × 𝑎𝐺 + 𝐻𝑉% × 𝑎𝐻𝑉 + 𝐼𝐷 × 𝑎𝐼 + × 𝑎𝑣𝑚 + × 𝑎𝑣𝑑
1,000 1,000
𝐿𝑠
+ × 𝑎𝐿𝑆 + 𝑉𝑅 × 𝑎𝑉𝑅
1,000
𝑎2 = 0.01797
𝑣𝑅,𝑚 𝑣𝑅,𝑑
𝑏2 = 𝑏0 + 𝐺 × 𝑏𝐺 + 𝐻𝑉% × 𝑏𝐻𝑉 + 𝐼𝐷 × 𝑏𝐼 + × 𝑏𝑣𝑚 + × 𝑏𝑣𝑑
1,000 1,000
𝐿𝑠
+ × 𝑏𝐿𝑆 + 𝑉𝑅 × 𝑏𝑉𝑅
1,000
𝑏2 = 0.2318
4,512
𝐿𝐹𝑅2 = 0.01797 × ln ( ) + 0.2318 = 0.224
7,148
The same procedure is applied to obtain the LFR for lane 3, using the
respective coefficients from Exhibit 38-C2:
𝑣𝑅,𝑚 𝑣𝑅,𝑑
𝑎3 = 𝑎0 + 𝐺 × 𝑎𝐺 + 𝐻𝑉% × 𝑎𝐻𝑉 + 𝐼𝐷 × 𝑎𝐼 + × 𝑎𝑣𝑚 + × 𝑎𝑣𝑑
1,000 1,000
𝐿𝑠
+ × 𝑎𝐿𝑆 + 𝑉𝑅 × 𝑎𝑉𝑅
1,000
𝑎3 = 0.09830
𝑣𝑅,𝑚 𝑣𝑅,𝑑
𝑏3 = 𝑏0 + 𝐺 × 𝑏𝐺 + 𝐻𝑉% × 𝑏𝐻𝑉 + 𝐼𝐷 × 𝑏𝐼 + × 𝑏𝑣𝑚 + × 𝑏𝑣𝑑
1,000 1,000
𝐿𝑠
+ × 𝑏𝐿𝑆 + 𝑉𝑅 × 𝑏𝑉𝑅
1,000
𝑏3 = 0.2992
4,512
𝐿𝐹𝑅3 = 0.09830 × ln ( ) + 0.2992 = 0.254
7,148
Finally, the LFR for the leftmost lane (lane 4) is obtained from Equation 38-C2:
𝐿𝐹𝑅4 = 1 − 𝐿𝐹𝑅3 − 𝐿𝐹𝑅2 − 𝐿𝐹𝑅1 = 1 − 0.254 − 0.224 − 0.261 = 0.261
It is worth noting that the methodology predicts flow distribution and
speeds for lanes upstream of the on-ramp, which means the weaving auxiliary
lane is not covered by the scope of this example problem.
Exhibit 38-C12
Comparison of Predicted and
Field-Measured Lane-by-Lane
Speeds
REFERENCES
C-1. University of Florida Transportation Institute; Cambridge Systematics, Some of these references can
be found in the Technical
Inc.; and A. Skabardonis. NCHRP Web-Only Document 290: Highway Reference Library in Volume 4.
Capacity Manual Methodologies for Corridors Involving Freeways and Surface
Streets. Transportation Research Board, Washington, D.C., 2020.
C-2. Sasahara, F., L. Elefteriadou, and S. Dong. Lane-by-Lane Analysis
Framework for Conducting Highway Capacity Analyses at Freeway
Segments. Transportation Research Record: Journal of the Transportation
Research Board, Vol. 2673, Issue 8, 2019, pp. 523–535.
C-3. Sasahara, F., L. Carvalho, T. Chowdhury, Z. Jerome, L. Elefteriadou, and A.
Skabardonis. Predicting Lane-by-Lane Flows and Speeds for Freeway
Segments. Transportation Research Record: Journal of the Transportation
Research Board, Vol. 2674, Issue 9, 2020, pp. 1052–1068.