COMPARISON OF FREE FLOW SPEED ESTIMATION MODELS

Usman Tasiu Abdurrahman1, Othman Che Puan2* Muttaka Na’iya Ibrahim3

1,2,3
Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 Skudai, Johor, Malaysia

Email: [email protected], [email protected], [email protected]

*Corresponding author: [email protected]

ABSTRACT
Free flow speed (FFS) is the drivers’ desired speed on roadways at low traffic volume and absence of traffic
control devices whose determination is a fundamental step in the analysis of two-lane highways. FFS can either
be estimated using either analytical model or based on field measurement. Regarding the former approach; the
Malaysian Highway Capacity Manual (MHCM) established a model for estimating FFS based on base-free-
flow-speed (BFFS), roadway’s geometric features and fraction of motorcycles in the traffic stream. On the other
hand, the Highway Capacity Manual (HCM) suggested an approach for field measurement of FFS; preferably at
a two-way flow rate not exceeding 200 veh/h. For many highways, observing a two-way flow rate of 200 veh/h
or less is seldom met or impossible. In such situations, mean speed could be observed at higher flow rates and
adjusted accordingly using a model provided by the HCM. This study describes the application of the two
approaches for measuring FFS. Moving car observer (MCO) method was used for collecting the relevant data
related to travel time, speed, flow rate, and traffic composition with using a video recording instrumented test
vehicle while the roadway geometric features were measured manually. Data for the study were collected on
four directional segments of rural two-lane highways with varying geometric features and traffic composition in
Johor, Malaysia. Field data obtained were analyzed to estimate the FFS using the two approaches. Results
obtained from both methods were compared to ascertain the degree of their consistency or otherwise. Statistical
analysis using student t-test indicates that there is no statistically significant difference between the FFS
estimates from the two approaches.

Keywords: Two-lane highways, Free flow speed, Estimation, Malaysian Highway Capacity Manual, HCM

1.0 Introduction

Free-flow speed (FFS) refers to an average speed of vehicles on road segments not close to an

intersection under conditions of low vehicular density. It is a significant variable used in

estimating the expected operating conditions of highways, and it is only possible when the

traffic volume on the road segment is below capacity. A key step in analyzing capacity and

level of service for uninterrupted flow condition is the determination of free-flow speed. FFS

together with demand flow rates are used in determining average travel speed of roadway

facility. It has been established that various factors relating to road geometry, visibility and

weather conditions influence FFS (Brilon and Ponzlet, 1996; Ibrahim and Hall, 1994; Kyte et

al., 2000; Medina and Tarko, 2005; TRB, 2010; Yagar and Van Aerde, 1983)

FFS can either be determined using direct field measurements or estimated using a model

(TRB, 2010). Regarding the direct field measurement, the Highway Capacity Manual, (HCM
2010) (TRB, 2010) recommended that FFS can be measured directly in the field at a two-way

flow arte not exceeding 200veh/h. According to the HCM, average running speed of the

stream under such flow rate limit can be reported as FFS. However, for conditions where the

flow rate exceeds 200veh/h, a model was established by the HCM to adjust the stream speed

into a FFS; provided that the data was based on direct field measurement.

For the indirect approach, Malaysian Highway Capacity Manual (MHCM) (HPU, 2011)

provides an analytical model for estimating FFS in which base-free-flow-speed (BFFS),

highway’s geometric features, and proportion of motorcycles were used as the model inputs.

In this study, FFS was evaluated on two-lane highways based on HCM and MHCM models at

two-way flow rates higher 200veh/h on same set of roadway segments. FFS estimates from

the two models were compared to ascertain their consistency or otherwise.

2.0 Experimentation

Data for this study were collected on four (4) directional segments of two-lane rural highways

drawn from Pontian – Kukup (PTN – KKP) and Renggam – Kulai (REN – KUL), Johor,

Malaysia. Data relating highways’ geometric features, speed, and flow rates were identified

as the major inputs for estimating FFS using both the HCM and MHCM models as the case

may be. Inputs regarding the roadways’ geometric features were measured manually using

measuring tape. Speed and flow rates related parameters were collected using moving car

observer (MCO) in accordance with the procedures described in the Manual of

Transportation Engineering Studies (Robbertson and Findley, 2010) based on floating-car

driving technique. In floating-car driving style, the test vehicle is driven into the traffic

stream under study and overtakes as many vehicles as overtaking it; through this, the test car

estimates the behaviour of an average vehicle in the traffic stream (Roger et al., 2004). The

speed of the test vehicle is thus regarded as the average speed of the traffic stream evaluated.
In applying the MCO method, a segment length of 3.50 km was used for the data collection

by performing six (6) test runs on each directional segment; as six runs were found to be

satisfactory for consistent and unbiased estimates of measured variables (Mortimer, 1957). A

passenger car equipped with video recording system was used as the test vehicle. The video

recording system captures real time traffic events over the entire period of the test runs and

stores the recorded traffic events onto an SD memory card inserted into the recorder and

subsequently uploaded to computer for processing. The recorded traffic events were then

played back in a computer to extract the required data. During the playback, the time taken to

traverse the study segment was noted while the numbers of vehicles against the test car travel

direction, vehicles overtaking the test car and vehicles passed by the test car were extracted

respectively. The hourly flow rates for northbound and southbound directions were

determined using equations (1) and (2), respectively.

60( M s  On  Pn )
Vn  (1)
(Ts  Tn )

60( M n  Os  Ps )
Vs  (2)
(Tn  Ts )
Where,
V = Directional hourly volume (veh/h)
M = Opposing vehicles to the test car’s direction of travel (veh)

O = Vehicles overtaking the test car (veh)

P = Vehicles passed by the test car (veh)
T = Directional travel time taken to traverse the study segment (minutes)

The subscripts n and s refer to northbound and southbound directions, respectively.

2.1 HCM Adjustment Model for FFS Estimation

Based on the HCM, for speed study conducted at a two-way flow rate of more than 200

veh/h, a volume adjustment must be made in order to determine FFS. Equation (3) was

provided by the HCM for making the adjustment.

Vf
FFS  SFM  0.00776 (3)
fHV

Where,

FFS = Estimated Free-Flow Speed (km/h)

SFM = Mean Speed of traffic measured in the field (km/h)

Vf = Observed flow rate for the period when field data were obtained (veh/h)

fHV = Heavy-vehicle adjustment factor, computed using equation (4)

1
fHV 
1  PT ( ET  1)  PR( ER  1) (4)

Where,

PT = Proportion of trucks in the traffic stream, expressed as a decimal

PR = Proportion of recreational vehicles (RVs) in the traffic stream, expressed as a decimal

ET = Passenger-car equivalent for trucks

ER = Passenger-car equivalent for RVs

Mean speed of traffic stream was obtained by taking the ratio of the segment length to the

travel time taken to traverse the study segment. Directional traffic flow rates and composition

of heavy traffic (trucks) was extracted from the field recorded data using the MCO method.

Heavy vehicles adjustment factors were obtained from tables provided by the HCM.
2.1 MHCM Model for FFS Estimation

The MHCM provided an FFS estimation model for two-lane highways, based on BFFS,

adjustment for the effect of lane and shoulder widths narrower than 3.65 m and 1.80 m

respectively as well as the effect of proportion of motorcycles in the traffic stream. Equation

(5) shows the MHCM model in which a BFFS value of 90 km/h was recommended.

FFS  BFFS  f LS  f APD  f m (5)

Where,

FFS = Free-Flow Speed (km/h)

BFFS = Base Free-Flow Speed (km/h)

fLS = Adjustment for lane and shoulder widths less than 3.65 m and 1.80 m, respectively

(km/h)

fAPD = Adjustment for access points density (km/h)

fm = Adjustment for proportion of motorcycles (km/h)

3.0 Results and Discussions

3.1 Geometry of the study segments

As stated earlier, four directional segments were chosen for this study. For each of the

segments, data were collected and reported as northbound (NB) or southbound (SB)

directions depending on the segments. Table 1 presents the geometric features of the

segments used in this study which are parts of the required inputs in estimating FFS using

MHCM model.

Table 1: Roadways Geometry

Road Direction Lw SHw APD
(m) (m) (access/km)
NB 3.09 0.25 1.71
PTN – KKP
SB 3.09 0.26 1.71
NB 3.65 1.50 0.29
REN - KUL
SB 3.65 1.60 0.29
Lw = Lane width, SHw = Shoulder width, APD Access point density
3.2 Free Flow Speed Estimation

Free flow speeds were estimated on the selected two-lane highway segments using both the

HCM and MHCM as described in the subsequent sections. FFS estimates from the two

models were also compared.

3.2.1 FFS Estimation Using HCM Adjustment Model

Equations (3) and (4) were used for the estimation of FFS in this case. Directional mean

speeds of the segments were first obtained using MCO after which the directional traffic

volumes were determined alongside with the proportion of trucks for each direction. Exhibits

presented in HCM were used for the determination of passenger car equivalents (PCE) for

trucks used in equation (4) for the determination of heavy-vehicle adjustment factor. Trucks

were the only type of heavy vehicle for Malaysian traffic condition, as such recreational

vehicles (RV) were not considered in the analysis. Table 2 shows the flow rates, mean

speeds, heavy-vehicle adjustment factors, as well as the estimated FFS for the four directional

segments.

Table 2: FFS Using HCM Adjustment Model

Travel Mean
q PT FFS
Road Direction time Speed PCE fHV
(veh/h) (km/h)
(min) (km/h)
NB 299 2.82 74.42 0.08 1.40 0.97 76.82
PTN – KKP
SB 195 2.74 76.60 0.03 1.50 0.98 78.13
NB 164 2.40 87.62 0.07 1.60 0.96 88.95
REN - KUL
SB 259 2.40 87.44 0.05 1.40 0.98 89.49
q = average directional flow rate, PT = proportion of trucks

3.2.2 FFS Estimation Using MHCM Adjustment Model

Equation (5) was used for the determination of the FFS and the estimates are as shown in

Table 3. All adjustment factors were obtained from tables provided by the MHCM.
 Table 3 FFS Using MHCM Estimation Model

q PMc fLS fAPD fm FFS

Road Direction (veh/h) (%) (km/h) (km/h) (km/h) (km/h)
NB 299 0.26 7.80 2.04 2.60 77.56
PTN – KKP
SB 195 0.09 7.80 2.04 1.70 78.46
NB 164 0.06 1.00 0.35 0.78 87.87
REN - KUL
SB 259 0.06 0.70 0.35 0.78 88.17
FFS  90  f LS  f APD  f m , PMc = Proportion of motorcycles

3.3 Comparison of the two FSS Estimation Models

Table 3 presents a summary of the FFS estimates obtained from HCM and MHCM models

for ease of comparison.

Table 4: Comparison of FFS Estimation Models

Road Direction FFSHCM FFSMHCM

(km/h) (km/h)
NB 76.82 77.56
PTN – KKP
SB 78.13 78.46
NB 88.95 87.87
REN - KUL
SB 89.49 88.17
Mean values (km/h) 83.35 83.02

FFS estimates from the two models seem to differ slightly. However, while for PTN – KKP

segment FFS values based on HCM model were slightly lower than those from MHCM

model; in the case of REN – KUL segment, an opposite of that recorded. This indicates that

estimates from the two approaches do not follow a particular pattern. Based on the irregular

trend shown by the FFS estimates from the two models, a more reasonable comparison could

be drawn using the mean FFS values. On the basis of the mean FFS values, the two models

resulted in consistent estimates even though, MHCM estimates were lower than HCM values

by about 0.4%. This difference could be deemed negligible enough that may not cause any

considerable effect. To ascertain the extent of the effect of this difference could cause if any,

a statistical analysis using t-test at 95% confidence level was carried out to find out whether

the difference between the means of the FFS is significant.

Results from the statistical analysis revealed that there is statistically significant difference

between the two sets of data as p- value (0.5610) is far greater than 0.05. This finding

suggests that either of the equations can be used to estimate FFS on two-lane highways.

4.0 Conclusion

Free-flow-speed is an essential parameter in the capacity and level of service analyses for

two-lane highways. It is estimated either through direct field measurement at two-way flow

rate not exceeding 200 veh/h or indirectly using analytical model. Direct field measurement

of FFS is seldom realistic as roadways mostly operate at flow rates higher than the specified

level. This made it necessary to utilize analytical models alongside with some adjustments to

estimate FFS. The current study presented an evaluation of FFS on two-lane highways based

on the indirect measuring approach using HCM and MHCM models. Results obtained from

the two approaches were compared to establish their consistency or otherwise. Findings from

statistical analysis using t-statics revealed that there is no statistically significant difference

between the two data sets as p – value (0.5610) is far greater than 0.05. A key implication of

this finding is that either of the models can be used for indirect estimate of FFS; especially

for situations where the combined directional flow rates exceed the minimum specified.

Acknowledgement

The authors would like to express deep appreciations to the Ministry of Higher Education,

Malaysia, through Research Management Centre (RMC), Universiti Teknologi Malaysia

(UTM) and UTM for providing a research grant (Q.J130000.7801.4L100), opportunity and

necessary facilities to support this research work.

References

Brilon, W. and Ponzlet, M. (1996). Variability of speed-flow relationships on German

autobahns. Transportation Research Record: Journal of the Transportation Research
Board. 1555(1), 91-98.
HPU (2011). Malaysian Highway Capacity Manual. Highway Planning Unit, Ministry of
Works, Malaysia.
Ibrahim, A. T. and Hall, F. L. (1994). Effect of adverse weather conditions on speed-flow-
occupancy relationships.
Kyte, M., Khatib, Z., Shannon, P. and Kitchener, F. (2000). Effect of environmental factors
on free-flow speed. US National Academy of Sciences Transportation Research
Board.
Medina, A. M. F. and Tarko, A. P. (2005). Speed factors on two-lane rural highways in free-
flow conditions. Transportation Research Record: Journal of the Transportation
Research Board. 1912(1), 39-46.
Mortimer, W. J. (1957). Moving Vehicle Method of Estimating Traffic Volumes and Speeds.
Robbertson, H. D. and Findley, D. J. (2010). Manual of Transportation Engineering Studies.
(2nd ed.)Institute of Transportation Engineers.
Roger, P. R., Elena, S. P. and William, R. S. (2004). Traffic Engineering, 3rd Ed. New
Jersey: Pearson Prentice Hall.
TRB (2010). Highway Capacity Manual, 5th Ed. National Research Council, Washington,
D.C.
Yagar, S. and Van Aerde, M. (1983). Geometric and environmental effects on speeds of 2-
lane highways. Transportation Research Part A: General. 17(4), 315-325.