Vous êtes sur la page 1sur 5

The Highway Capacity

Manual Delay Formula for


Signalized Intersections
BY RAHMI AKCELIK

T he purpose of this article is to com-


pare the 1985 U.S. Highway Capac-
ity Manual (HCM) delay formula for
forgotten and the formula misused in
practice (e.g., in evaluating alternative
designs or in stating benefits from im-
c
u
= signal cycle time in seconds,
= g/c (ratio of effective green time
to cycle time),
signalized intersections with the provements to an existing oversaturated x = degree of saturation (ratio of ar-
Australian and Canadian3 formulas and intersection). rival flow rate to capacity),
to present a generalized form that em- flow period in hours,
braces them all. The aim is to promote ;: capacity in vehicles per hour,
international cooperation in this area of The HCM formula m,n = calibration parameters, and

I
research and development. predicts higher X. = the degree of saturation below
Compared with the other delay for- which the second term of the de-
mulas, the HCM formula predicts higher delays for lay formula is zero. This can be
delays for oversaturated conditions, and oversaturted expressed as
the differences between the prediction x = a + bsg (2)
from the HCM formula and the other conditions. where:
formulas increase with increasing degree
of saturation. An alternative to the The generalized formula could be cal- Sg = capacity per cycle (s = satura-
HCM delay formula, which matches the ibrated to develop a more suitable for- tion flow rate in vehicles per sec-
other formulas for oversaturated condi- mula for U.S. conditions. It might also ond and g = effective green
tions, is given. The alternative formula, be possible to calibrate it for vehicle-ac- time in seconds), and
derived from the generalized formula, tuated and fixed-time signals separately. a,b = calibration parameters.
predicts delays that are very close to Various other issues related to this dis- The two terms of the delay formula
those predicted by the original HCM cussion and briefly mentioned in this ar- can be referred to as the uniform delay
formula for undersaturated conditions, ticle will be discussed elsewhere. (du) and the overflow (dO) terms:
and at the same time predicts delays that d=du+ do. (3)
are very close to the results from the
Australian and Canadian formulas for
A Generalized Formula The overflow delay is called the incre-
oversaturated conditions. The HCM, Australian, and Canadian mental delay because of random arrivals
The HCM signalized intersection formulas for delay at traffic signals can and individual cycle failures in the
chapter states that its delay formula be generalized as the following two-term HCM. The usefulness of this concept is
yields reasonable results for values of .x equation: that an overflow queue formulation can
between 0.0 and 1.0. . . The equation be used as a common base for the for-
~ = 0.5C(1 u)
may be used with caution for values of x + 900 TX (x-1) mulas to predict delay, number of stops,
1UX [ and queue length as in the Australian
up to 1.2, but delay estimates for higher
values are not recommended. Although
traffic engineers do not design for over-
saturation, a delay formula that can be
+

where:
v (x- 1~ + m(x-xo)/QT
J (1)
method. 2 The relation between the av-
erage overflow queue (No) in vehicles
and the overflow delay (dO) in seconds is
expected to give reasonable results for 3600 do = NOIQ (4)
oversaturated, as well as undersaturated d = average overall delay (including
stop-start delays) in seconds per where Q is the capacity in veh.lhr. In this
conditions, is preferred because the lim-
vehicle, sense, Equation 1 is based on a gener-
itations of this type of formula are often

ITE JOURNAL . MARCH 1988023


alized overflow queue formula. More de- Table 1. Values of the Calibration Parameters in the HCM, Australian, and Canadian
tailed discussion on the use of the over- Overflow Deiay Formuias, Expressed in Terms of Equation 1
flow queue concept to predict delay,
n m a b
number of stops, and queue length is
given in the Appendix. HCM 2 4 0 0
Various specific delay formulas can be Australian o 12 0.67 1/600
Canadian o 4 0 0
derived from Equation 1 by setting the
TRANSW 8 1 4 0 0
calibration parameters n, m, a, and b in
Alternative
the overflow delay term to appropriate to HCM o 8 0.50 0
values. Therefore, only the second term
of Equation 1 will differ between alter-
native models as no calibration parame- Tabie 2. Overflow Delays in 3econds per Vehicie from the HCM, Australian, and
ters are considered in the first term (dJ. Canadian Delay Formulas (c = 90 sec., g = 30 sec., and Q = 500 veh./ho

The values of the calibration parameters x HCM Alternative Australian Canadian Deterministicb
n, m, a, and b for the 1985 HCM , 1981
o 0.0 0.0 0.0 0.0
Australian, and 1984 Canadian,3
0.4 0.4 0.0 0.0 2.4
models are given in Table 1. The HCM
0.6 1.9 1.8 0.0 5.2
model differs from the other two models 0.8 8.4 9.7 5.5 12.6
with the x factor (n = 2) whereas the 0.9 17,6 19.9 16.5 21.8
Australian model differs from the other 0.95 26.6 28.5 25.9 29.5
two models with a nonzero XOparameter 1.0 40.2 40.2 38.8 40.2 0.0
(X. = 0.67 + sg/600). 1.1 85.4 72.0 72.4 70.3 45.0
The exact form of the HCM formula 1.2 155.5 110.5 112.1 108,0 90.0
differs from Equation 1 because a factor 1.4 376.0 195.0 197.5 191.8 180.0
of 1/1.3 = 0.77 is applied to convert the WVhererr =0, rn=8, c7=0.5, andb= O.
overall delay to stopped delay (i.e., it bFrom Equation 60,

assumes that the stopped delay is always


779%of the overall delay). Whilst the au- 160

thor does not necessarily agree with this


simplifying assumption, it is outside the
t _ USHCM
scope of this article. 140
To facilitate comparisons, T = 0.25 _ Australian

hr., which is fixed in the HCM model,


will be used for all models in the follow-
t

I .._- Canadian

120 . Deterministic
ing numerical example although it is best
to leave this as a variable in the general
model. It will also be assumed, by way
of example, that c = 90 sec., g = 30 t
sec., s = 1500 veh.lhr., and therefore gl Overflow
c = 1/3, Q = 500 veh.lhr., and sg = de~y,

12.5 veh. The overflow delay, do, from (secon;s per


vehicle)
the second term of Equation 1 for this
example is given by
do = 225 X [(X 1)
+ V(X - 1~ + m(x-xO)/125] (5)
for x > x,, (O otherwise)
The results from Equation 5 for the
HCM, Australian, and Canadian models
are shown in Figure 1, and tabulated in
Table 2. For this example, x. = 0.691 for
the Australian model and x. = Ofor the
other two. 20
I /
Time-Dependent Delay
Formulation
Itis seen from Figure 1 that the Austra-
o~
0.2
----.---:---//
0.4
/d
0.6 0.8 1.0 1.2
Degree of saturation, x
lian and Canadian models produce ov-
erflow delay curves that are asymptotic Figure 1. Overtlow delays predicted by the HCM,Australian,andCanadianformulas
to a deterministic oversaturation delay (c= 90 sec., g=30 sec., Q=500 veh.hw., and T= 0.25 hr.).

24. iTE JOURNAL . MARCH 1988


line (Akcelik) given by well-known Webster formula. 7The Aus- by setting n = O (i.e., deleting the x
d = 1800T(x 1) for .x=1 tralian formula corresponds to k = 1.5 factor and choosing the appropriate val-
(6)
with a variable value of XO,which is an ues of parameters m, a, and b). It has
For the example above, approximation4 to Millers delay been found by a best-fit analysis of the
d = 450 (xl). (6a) formula. Unlike the Webster formula, overflow delay values from the HCM for-
Millers original equation was based on mula for x values in the range 0.6 to 1.0
This deterministic delay formula is
the formulation of overflow queues. This that the parameter values of m = 8 and
based on a single flow period of length
was extended to oversaturated condi- a = 0.50 give a satisfactory solution
T with constant flow and capacity and
tions by the author using the TRRL when parameter b is set to zero to obtain
with no initial overflow queue. The delay
time-dependent delay method and intro- a formula with constant Xo. Using T =
from Equations 6 or 6a includes delay
ducing a simplification to the original 0.25 hr., this overflow delay formula is
during T as well as delay after T (as ex-
Miller formula.4 The reader is also re- expressed as
perienced by the vehicles that arrive dur-
ferred to a paper by Hurdle, which dis- dO = 225 [(X 1)
ing T but may depart after T). Subse-
cusses the relationship between the
quently, the generalized formula, + V(X- 1)2 + 32 [(x- O.5)/Q] (8)
steady-state and time-dependent delay
Equation 1, is based on the same simpli-
formulas. for x >0.5 (zero otherwise).
fying assumption, which is used as an
As seen in Figure 1, the HCM formula This equation gives the same delay value
alternative to the more complicated var-
appears to produce a curve that does not as the HCM equation for flow at capacity
iable-demand analysis methods that
have the fundamental characteristic of (x = 1.0). Interestingly, it corresponds
treat the peaking of arrival flows explic-
the time-dependent delay formulation. to an earlier steady-state overflow delay
itly.
For x above 1.0, it diverges from the de- formula given by Miller, which approx-
The asymptotic curve is an important
terministic delay line and predicts very imates to Equation 7 with k = 1.0 and
characteristic of time-dependent delay
large delay values as indicated by the x = 0.5. The overflow delay results
formulation and was originally devel-
shaded area. This is due to the X2factor from Equation 8 for Q = 500 veh./hr.
oped by researchers at the U.K. Trans-
(n = 2 in Equation 1). It is not clear if are given in Table 2 as the alternative
port and Road Research Laboratory for
the X2factor was introduced in an effort model.
the TRANSYT programs. In fact, the
to calibrate the delay formula for under- For direct comparison with the full
Canadian delay formula is the same as
saturated conditions, or for reasons re- HCM formula, the stopped delay for-
the formula given by Robertson. How-
lated to the use of a fixed value of T = mula obtained by replacing the second
ever, TRANSYT Version 8 uses a differ-
15 minutes (peak flow period). A clari- term of Equation 1 by Equation 8 and
ent form of the function equivalent to
~=l, m=4, a=O, and b=Oin fication of this issue would explain a fun- applying the factor of 0.77 to both terms
damental difference between the HCM is as follows:
Equation 1. Different forms of the func-
formula and the other formulas.
tion were used in earlier versions of 0.385c(1 u)
TRANSYT. An Alternative to the HCM d, = + 173 [(x 1)
1UX
Formula + V(X - 1)2 + 32 (x- O.5)/Q (9)
Converting the
A formula that gives delay values close The results from Equation 9 and the
overall delay to to the HCM formula for x values less HCM formula for stopped delay for the
speed delay needs than 1.0, but remains asymptotic to the above example (c= 90 sec., g = 30 sec.,
deterministic oversaturation line (Equa- and Q = 500 veh./hr. ) are given in Table
particular tion 6), can be derived from Equation 1 3. The values in Table 3 for x larger than
attention before
Table 3. Comparison of Sopped Deiavs from the HCM Formula and the
any calibration Recommended Alternative Formulation Using Factor of 0.77 (c= 90 sec., g= 30 sec.,
effort. and Q= 500 veh./hr.)

The time-dependent delay models are HCM Alternative Difference


derived by converting a steady-state de- x (sec.] (8ec.] (sec.] Y. Differenced
lay function, which is applicable to un- 0.2 16.5 16.5 0 0
dersaturated conditions only, to an 0.4 18.0 17.8 0.2 1,1
asymptotic time-dependent function, 0.6 20.7 20.6 0.1 0.5
which becomes applicable to oversatur- 0.8 27.2 28.5 +1.3 + 4.8
ated conditions also. The steady-state 0.9 35.6 37.3 +1.7 + 4.8
function that corresponds to Equation 1 0.95 43.0 44.5 +1.5 + 3.5
1.0 54.0 54.0 0 0
when n = O can be expressed as
1.1 88.5 78.5 10.0 11,3
k(x x<,)x/(l X) (7) 1.2 442.7 108.1 34.6 24.2
1.3 216.9 140.0 -76.9 -35.5
where parameter k is related to param-
1,4 312.3 173.0 -139.3 -44.6
eter m in Equation 1 by m = 8k. The
Canadian formula corresponds to %Yhere n=2, m=4, a=O, and/J=O.
bWheren=O, m=8, a=0,5, and b=0.
Equation 7 where k = 0.5 and X. = O, C(Alternative - HCM).
which is the random delay term of the 100 x (Alternative HCM)/HCM.

ITE JOURNAL. MARCH 1988.25


1.0 have been calculated using the value HMC formula for degrees of saturation form of the model has certain advan-
of the uniform delay at capacity (first less than 1.0, and at the same time is tages.
term for x=1) as a fixed value (d= 23.1 similar to the Australian, Canadian, and Furthermore, the question of convert-
sec. ) for both models .Explanation of TRANSYT formulas in producing a de- ing the overall delay to stopped delay
this method used by the SIDRA com- lay curve asymptotic to the deterministic needs particular attention before any
puter program- can be found in an delay line for degrees of saturation calibration effort. Researchers could
earlier article by the author. The maxi- greater than 1.0. also try deriving separate formulas for
mum difference between the HCM for- However, the calibration of the gen- fixed-time and vehicle-actuated signals
mula and the alternative formula for x eralized formula (Equation 1) directly by calibrating the generalized Equation
less than 1,0 is about 2 sec. (59%). using actual data for U.S. conditions 1 of this article (n = O recommended).
rather than the results from the original Again, the important question of apply-
HCM formula would, of course, be a ing the delay formulas for individual
Conclusion better way of developing an alternative lanes as against lane groups has not been
Equation 9 should be considered as an formula. It would also be useful to seek discussed here. In this respect, any cal-
alternative to the HCM delay formula. a value of X. dependent on capacity per ibration effort should take the particular
This formula gives values close to the cycle (sg) in this process, because this method of application into account.

26. ITE JOURNAL . MARCH 1988


Acknowledgment 3. Teply, S, (cd.). Canadian Capacity Guide 6. Vincent, R, A.; Mitchell, A. 1.; and Rob-
The author thanks the executive director for Signalised Edmonton, Al-
Intersections. ertson, D. 1. User Guide to TRANSYT Ver-
berta, Canada: Institute of Transportation sion 8. TRRL Laboratory Report LR 888.
of the Australian Road Research Board,
EngineersDistrict 7 and the University of United Kingdom: Transportation Road Re-
Dr. M. G. Lay, for permission to publish Alberta, 1984. search Laboratory, 1980.
this article. 4, Akcelik, R, Time-Dependent Expressions 7. Webster, F. Y Trafjc Signal Settings. Road
for Delay, Stop Rate and Queue Length at Research Laboratory Technical Paper No. 39.
Traf/ic Signals. Internal Report AIR 367-1. London, England: HMSO, 1958.
References Nunawading, Australia: Australian Road Re- 8. Miller, A. J, Australian Road Capacity
1. Transportation Research Board, National search Board, 1980. GuideProvisional Introduction and Signal-
Research Council. Highway Capacity Man- 5. Robertson, D. I. Traffic Models and Op- ised Intersections. Bulletin No. 4, 1968. (Re-
ual. Special Report 209. Washington, DC: timum Strategies of Control: A Review. In: printed as ARRB Research Report No. 79,
Transportation Research Board, 1985. Proceedings of the International Symposium 1978. Replaced by ARR No. 123, 1981). Nun-
2. Akcelik, R. Traffic Signals: Capacity and on Traffic Control Systems, vol. 1. University awading, Australia: Australian Road Re-
Timing Analysis. Research Report ARR No. of California, Berkeley, California: Institute search Board, 1981.
123. Nunawading, Australia: Australian of Transportation Studies and the U.S. De- 9. Hurdle, V F. Signalised Intersection Delay
Road Research Board, 1981. (Third reprint partment of Transportation, 1979, pp. 262- ModelsA Primer for the Uninitiated. Trans-
1986). 288. portation Research Record 971. Washington,
DC: Transportation Research Board, 1984,
pp. 96-104.
10. Miller A. J. Settings for Fixed-Cycle
Traffic Signals. Operations Research Quar-
terly 14(4): 373-386 (1963).
11. Akcelik, R. SIDRA-2 does it Lane by
Lane. 1ss Proceedings of the 12th ARRB
Conference 12(4): 137-149 (1984),
12. Akcelik, R. Sidra Version 2.2 Input and
Output. Australian Road Research Board.
Technical Manual ATM No. 19, 1986.
13. Akcelik, R. Introduction to SIDRA-2 for
Signal Design: 13th ARRB Conference Work-
shop Papers and Discussions. Australian
Road Research Board. Research Report
ARR No. 148, 1987.
14. Akcelik, R. Estimating the Capacity of
a Shared Lane. Presented at the Third Mini
Euro Conf. on Operations Research Methods
in Transport Planning and Traffic Control.
Herceg Novi, Yugoslavia, June 1987.
15. Powell, J.L. Project evaluation at sig-
nalised intersections using the 1985 HCM.
In Compendium of Technical Papers. Wash-
ington, DC: Institute of Transportation En-
gineers, pp. 81-85, 1987. I

Rahmi Akcelik is
a principal re-
search scientist at
the Australian
Road Research
Board. His current
work is in the traf-
jic signal and fuel consumption areas of
research and development. He is a mem-
ber of ITE Technical Committee 5B-27j
Left-Turn Lane Storage Length Design
Criteria. He graduated from Istanbul
Technical University, Turkey, in 1968 and
received his Ph. D. degree in transporta-
tion engineering from the University of
Leeds, England, in 1974. He is a Member
of ITE.

ITE JOURNAL . MARCH 1988.27

Vous aimerez peut-être aussi