Académique Documents
Professionnel Documents
Culture Documents
1
2
3 AN ADVANCED SIGNAL PHASING SCHEME FOR DIVERGING DIAMOND
4 INTERCHANGES
5
6
7 Peifeng Hu, Ph.D.
8 (Corresponding Author)
9 Department of Civil & Environmental Engineering
10 University of Nevada, Reno
11 Reno, NV, USA 89557
12 Email: hupeifeng1981@gmail.com
13 Tel: (775) 784-1232
14 Fax: (775) 784-1390
15
16 Zong Z. Tian, Ph.D., P.E.
17 Department of Civil & Environmental Engineering
18 University of Nevada, Reno
19 Reno, NV, USA 89557
20 Email: zongt@unr.edu
21 Tel: (775) 784-1232
22 Fax: (775) 784-1390
23
24 Hao Xu, Ph.D.
25 Department of Civil & Environmental Engineering
26 University of Nevada, Reno
27 Reno, NV, USA 89557
28 Email: haox@unr.edu
29 Tel: (775) 784-6909
30 Fax: (775) 784-1390
31
32 Rasool Andalibian, Ph.D. Candidate
33 Department of Civil & Environmental Engineering
34 University of Nevada, Reno
35 Reno, NV, USA 89557
36 Email: randalibian@gmail.com
37 Tel: (775) 784-6195
38 Fax: (775) 784-1390
39
40
41 Nov 2013
42
43 ABSTRACT
44 A diverging diamond interchange (DDI), also called a double crossover diamond interchange
45 (DCD), is a relatively new interchange type in the U.S. Due to its many advantages over the
46 conventional diamond interchanges such as higher capacity and lower delay, there has been a
47 growing interest in its analysis methodologies and operational strategies. DDIs have shown
48 particularly high efficiency under conditions with heavy ramp traffic to arterial roads. However,
49 as the DDI concept is relatively new in the U.S., the research on DDIs is still considered to be in
50 the preliminary stage. In particular, most studies found in the literature only provided basic
51 phasing schemes, which were not optimized for efficiency. Furthermore, there is a lack of
52 commonly acceptable methodologies for obtaining critical signal control parameters, such as the
53 cycle length, and phasing splits. Therefore, it is an area topic in need of much research in order
54 to develop signal phasing schemes for achieving the optimized DDI performance. This study
55 proposed a new methodology for signal phasing scheme design at DDIs. With minor adjustment,
56 it can be applied to a variety of DDIs with different traffic demands and geometric
57 configurations. The proposed phasing design approach was specifically designed for DDIs, so it
58 could be employed as a guide or a reference for signal timing design at DDIs. At the DDI of
59 Moana Ln and U.S. 395 in Reno, Nevada, the proposed scheme was evaluated and compared
60 with various signal phasing schemes by using the hardware-in-the-loop simulation technology.
61 The simulation results indicated that the proposed approach significantly outperformed the one
62 currently being used at this interchange.
63 Keywords: Diverging Diamond Interchange, Traffic Signal Control, Signal Phasing Scheme,
64 Webster Method, VISSIM.
65 INTRODUCTION
66 The DDI concept was first introduced to the United States by Gilbert Chlewicki in 2003 (1).
67 The first DDI was constructed in Springfield, Missouri in 2009 (2). Before that, the only known
68 DDIs existed in the communities of Versailles, Le Perreux-sur-Marne, and Seclin in France (3).
69 Field and simulation studies have demonstrated many advantages of DDIs over the conventional
70 diamond interchanges. As a result, seven states (Missouri, Utah, Georgia, Kentucky, Maryland,
71 New York, and Nevada) have constructed DDIs with demonstrated operational improvements
72 by the end of year 2012. More than 17 states are currently constructing DDIs in the U.S. (3).
73 Studies have been conducted to evaluate the performance of DDIs (1, 4-6). Most of
74 these studies compared the performance of DDIs to conventional diamond interchanges and
75 single point urban interchanges. Researchers sought signal timing plans either manually or by
76 using Synchro to optimize signal timing for each traffic scenario. Since Synchro does not
77 provide the specific function of signal timing optimization at DDIs, some of the research
78 interpreted the two crossover intersections of DDIs as two separate intersections operated by
79 two controllers (2). The microscopic simulation software of VISSM was then employed to
80 evaluate the performance. The existing research results have indicated that DDI works better
81 than the conventional interchange designs in terms of delay, number of stops, queue length, and
82 capacity (2, 6).
83 FHWA’s research suggested that the DDI signal timing could be two-phase control, with
84 each phase dedicated for the alternative opposing movements (7). Signals at a DDI can be fixed
85 time and may be fully actuated to minimize delay. Detectors can be used on all approaches of
86 both crossovers, and durations of signal phases may vary on a cycle-by-cycle basis. A DDI can
87 be operated by one controller or two. FHWA also indicated a three-phase signal operation by
88 adding a dummy phase as the spacing between the crossover junctions is very close. Until now,
89 there is no commonly accepted signal timing methodology specifically for DDIs. Therefore,
90 professionals rely on personal experience or conventional approaches for other types of
91 interchanges.
92 This study proposed a new methodology for the signal phasing scheme design at DDIs.
93 With minor adjustment, it can be applied to a variety of DDIs with different traffic demands and
94 geometric configurations. The proposed phasing design approach was specifically designed for
95 DDIs, so it could be employed as a guide or a reference for signal timing design at DDIs. This
96 research paper consists of four major sections. Section 1 provides a comprehensive literature
97 review pertinent to DDIs. Section 2 briefly introduces the current operation scheme and presents
98 a mathematical model for traffic signal timing design of DDIs based on Webster’s method.
99 Section 3 compares performance of the current traffic signal operation scheme and signal timing
100 by the proposed methodology with simulation results at the DDI of Moana Ln and U.S. 395
101 interchange in Reno, NV. The results indicated that the performance of the proposed scheme is
102 better than the current operation scheme at the studied interchange under the traffic demand
103 conditions. The last section offers conclusions of this study.
104 METHODOLOGY
106 The proposed phasing scheme is for one controller operation and can be used with all control
107 types: pre-timed, semi-actuated, and fully actuated. It is a general phasing scheme design
108 methodology for DDIs. The proposed phasing scheme is shown in Figure 1 and Figure 2. The
109 proposed operation operates very efficiently by applying overlapping phases extensively.
123 between these two nodes. Although Phase 8 is not assigned to any of the movements, the phase
124 with a “Min Recall” ensures phases 3 and 7 not to come on at the same time as they are actually
125 conflicting phases. Without phase 8, phase 7 will extend to run simultaneously with phase 3. If
126 phase 8 is set up with “Max Recall, the signal operation the DDI is not efficiently as phase 3
127 cannot gap out when its pertinent traffic is low. Figure 2 shows the locations of phases in the
128 proposed DDI phasing scheme. The SB on-ramp right-turn traffic and the NB on-ramp right-
129 turn traffic are controlled by “yield” type in this study. They can also be controlled by overlap
130 phases such as phase 4 for SB on-ramp right-turn traffic and overlapping phases 1 and 3 for NB
131 on-ramp right-turn traffic. Therefore, the proposed phasing scheme can be easily adjusted for
132 controlling other DDIs with different configurations.
133
134 Figure 1 Phases, Rings, and Barrier for Proposed Operation
135
136 Figure 2 Phase Location Diagram for Proposed Operation
137
138 Figure 3 Critical Nodes of a DDI
141 Similarly to Webster’s method, the proposed operation assumes that the v/c ratios for critical
142 lane groups are equal, and that the green times allocated to critical lane groups are assumed to
143 be proportional to their saturation flow rates. Unlike the traditional Webster’s method, the
144 principle of deriving a real cycle length and the real splits (the cycle length and the splits shown
145 in a controller) of a DDI by the proposed operation is adding a certain of additional times
146 (coming from overlapping phases) to the real cycle length for calculating its effective splits in
147 proportion, and then deducting the additional times from the effective cycle length and splits to
148 obtain the real cycle length and the real splits, which will be input into a controller.
149 The proposed method introduces phases 2 and 6 into the signal operation, which lead to
150 two different timing schemes: phase 6 is one of the critical phases or phase 2 is the one of the
151 critical phases. The criterion for selecting the critical phases between phases 2 and 6 is
152 determined by the traffic demands and configurations of a DDI. The performance of these two
153 schemes is different as their operation efficiencies are different. The flow chart in Figure 4
154 indicates the basic steps for deciding which scheme to use.
155
156 Figure 4 Overview of Proposed Operation
157 2. Signal Timing Scheme 1 of Proposed Operation
158 Signal timing scheme 1 is for the situation that phases 6, 4, and 3 are critical phases. Phase 1 is
159 predetermined by the travel time of ,, sec.
161 For the traffic signal operation shown in Figure 1, the DDI traffic signal timing parameters must
162 satisfy the relationship in the following equation:
163
,, ,,
2 ∗ ,, (1)
164 where
165 : effective green time of phase 3 (s);
166 : effective green time of phase 4 (s);
167
: effective green time of phase 6 (s);
168
: sum of splits of phases 3, 4, and 6 shown in Figure 1 (s);
169 ,,: split of phase 1, fixed and determined by the travel time between the two signals
170 “11” and “7” through nodes “11,” “12,” and “7” shown in Figure 3; and
171 : total lost time per cycle (s);
172 For this kind of DDI, the effective cycle length ( ) for the three critical lane groups
173 (SBL, eastbound (EB), and westbound (WB)) satisfy the following relationship with
:
174
2 ∗ ,, (2)
175 Replacing , ,
and
in Equations (1) and (2) yields:
176 ∑
∗ (3)
177 Then, based on the assumption, for ≠ " ,the effective cycle length is:
# # #
178 % ∑'
* (4)
∑'
& ) $
$ $ (
(
(
179 where
180 + : flow ratio for critical lane group + ; and
181 ,: sum of all critical lane groups.
182 For this traffic signal timing plan, the total lost time is:
183 - - -
(5)
184 - -, -., -, , /0 1 (6)
185 where
186 - : lost time of critical lane group ( 4 345 6) (s);
187 -, : start-up lost time of critical lane group ( 4 345 6) (s);
188 -., : clearance lost time of critical lane group ( 4 345 6) (s);
189 , : yellow interval of critical lane group ( 4 345 6) (s);
190 /0 : all-red interval of critical lane group ( 4 345 6) (s); and
191 1 : extension of effective green time of critical lane group ( 4 345 6) (s).
192 - -, 2 (7)
193 The lost time of phase 3 is 2 sec, since it is overlapped by phase 1. Phase 1 is overlapped
194 by phase 2. Therefore, the westbound through traffic has start-up lost time of 2 sec as shown in
195 Equation (7).
196 The real cycle length is computed by:
#
197 ,, * ,, (8)
$
(
201 where
202 7 : effective green time for phase ( 3, 4, 6).
203 The actual green time shown in Figure 1 can be obtained by:
204 ∅ , /0 (10)
205 7 ∅ (11)
206 7 (12)
207
7 ∅ (13)
208 where
209 ∅ : is equal to ,, (s).
210 The phase splits 3, 4, and 6 are:
211 ∅ , /0 (14)
212 The + in Equation (4) can be replaced by:
213 + =3{ ?,@A,B , ?,C,D } (15)
?,@A,B ?,C,D
214 + =3{ A,@,F , A,@G,H } (16)
A,@,F A,@G,H
215 where
216 I, , : traffic volumes of lane group between nodes “14” and “7” (veh/h);
217 J, ,: saturation flow rate of lane group between nodes “14” and “7” (veh/h);
218 I,K,
: traffic volumes of lane group between nodes “5” and “6” (veh/h);
219 J,K,
: saturation flow rate of lane group between nodes “5” and “6” (veh/h);
220 I ,,L: traffic volumes of lane group between nodes “1” and “8” (veh/h);
221 J ,,L: saturation flow rate of lane group between nodes “1” and “8” (veh/h);
222 I ,
,M: traffic volumes of lane group between nodes “10” and “9” (veh/h); and
223 J ,
,M: saturation flow rate of lane group between nodes “10” and “9” (veh/h).
225 Signal timing scheme 2 is for the situation that phases 2, 3, and 4 are critical phases. In this
226 scheme, phase 1 is also predetermined by the travel time of ,, sec. Phase 1 serves SBL off-
227 ramp traffic with phase 6. Phase 1 also serves the WBT through traffic. Phase 7 is neglected in
228 the scheme 2 for studying the signal timing as phase 2 is the critical phase. It is not efficient to
229 calculate the cycle length and splits by adding phase 7 into the effective green time as phase 2
230 itself can serve NBL off-ramp traffic well, or in other words, phase 7 is not effective for the
231 entire intersection although it can still add an additional green time to phase 2. All the symbols
232 of the signal timing scheme 2 are same as the signal timing scheme 1.
233 1) Cycle Length
234 For the traffic signal operation shown in Figure 2, the DDI traffic signal timing parameters
235 should satisfy the relationship in Equation 17:
236
,,
,, (17)
237 Under this condition, the effective cycle length for the three critical lane groups (NBL,
238 EB, and WB) satisfy the following relationship with
:
239
,, (18)
240 Replacing , , and
in Equations (17) and (18) yields:
241 ∑
∗ (19)
242 Then, based on the assumption, for ≠ " , the effective cycle length is:
# # #
243 % ∑'
* (20)
∑'
& ) $
$ $ (
(
(
246 For this traffic signal timing plan, the total lost time is:
247 - - - (22)
248 - -, -., -, , /0 1 (23)
249 - -, 2 (24)
250 The lost time of phase 3 is only 2 sec, since it is overlapped by phase 1. Phase 1 is
251 overlapped by phase 2. Therefore, the westbound through traffic has start-up lost time of 2 sec
252 as shown in Equation (24).
253 2) Effective Green Times and Phase Splits
256 The actual green time illustrated in Figure 1 can be obtained by:
257 ∅ , /0 (26)
258
7 (27)
259 7 ∅ (28)
260 7 (29)
266 The interchange of Moana Lane and U.S. 395 is a diverging diamond interchange in Reno, NV,
267 as shown in Figure 5. The interchange has high right-turn traffic demands on the off-ramps. No-
268 turn-on-red is used on the southbound off-ramp due to the dual-lane geometry. Other peak hour
269 volumes of the interchange are shown in Figure 5.
E Moana Ln
Year 2015
AM (PM)
Peak Hour Volumes
AM Peak: 7:00-8:00
PM Peak: 17:00-18:00
270
271 Figure 5 Bird’s-Eye View and Peak Hour Volumes of the Moana Lane/U.S. 395
272 Interchange (8, 9)
274 The current actuated-coordinated signal timing plans running at the studied interchange were
275 developed by the City of Reno in 2012, with consideration of the heavy right-turn movements
276 on the off-ramps and southbound no-turn-on-red. The current signal phasing scheme has been
277 evaluated by Hu et al. in September, 2012 (8). The existing timing plan is displayed in Figure 6
278 and Figure 7.
279
280
281 Figure 6 Phase, Ring, and Barrier Diagram by City of Reno
282
283
284 Figure 7 Phase Designation Diagram by City of Reno
285 Phases 4 and 1 are not conflicting phases. There is no accurate way to determine the phase 1’s
286 split. Therefore, there is no accurate quantitative method to derive the cycle length for the
287 current method as well. The cycle lengths of the actuated-coordinated signal timing plans at the
288 DDI are provided as the existing actuated-coordinated plans for the adjacent signals: 110-sec
289 cycle for AM and 130-sec cycle for PM. In this study, the phase 1 split is assumed to be 10% of
290 the cycle length. Phases 2, 3, and 4 share the remaining time in proportion to their critical
291 saturation flow ratios. Table 1 shows each phase split in year 2015 AM and PM peak hours.
292 Table 2 provides the yellow and all-red intervals of the two actuated-coordinated plans for
293 current operation used in VISSIM simulation models.
294
295
296
297 Table 1 Cycle Lengths and Phase Splits of Current and Proposed Operations
Current Operation Proposed Operation
2015 AM 2015 PM 2015 AM 2015 PM
Cycle (sec) 110 130 110 130
Phase 1 (sec) 11 13 10 10
Phase 2 (sec) 23 24 30 29
Phase 3 (sec) 23 34 17 32
Phase 4 (sec) 53 59 53 59
Phase 5 (sec) N.A. N.A. 10 10
Phase 6 (sec) 23 24 30 29
Phase 7 (sec) N.A. N.A. 12 12
Phase 8 (sec) N.A. N.A. 58 79
298
299 Table 2 Clearance Intervals Used in VISSIM Simulation Models of Current Operation
Operation Phase 1 2 3 4 5 6 7 8
Critical Movement 10->9 3->8 3->8 12->7 None 12->7 None None
Current
Yellow (sec) 3.5 3.5 3.5 3.5 3.5
Operation
Red (sec) 1.5 3.5 3.5 2.5 2.5
Operation Phase 1 2 3 4 5 6 7 8
Critical Movement 3->8 3->8 None 12->7 None 10->11 5->4 None
Proposed
Yellow (sec) 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3
Operation
Red (sec) 3.5 3.5 2.5 2.5 1.5 1.5 1.5 0
355 CONCLUSION
356 Since there is no widely accepted signal timing methodology for DDIs, in the literature, traffic
357 engineers and researchers designed DDI signal timing plans based on their personal experience
358 and approaches for conventional interchanges, which have been proven to be inefficient. Some
359 of the signal phasing schemes in existing research did not consider the capability of signal
360 controllers, so they are not readily implementable in the field.
361 Because DDIs are a relatively new interchange design, there is a limited amount of related
362 literature. Most DDI related reports do not include detailed descriptions of feasible signal
363 phasing schemes. Where signal phasing was actually discussed, most reports did not follow the
364 NEMA convention, thus the phasing schemes were not readily implementable in the field. This
365 paper presented an innovative phasing scheme and a methodology for determining cycle length,
366 splits, and phase intervals for DDIs with consideration of the capacity of signal controllers. The
367 proposed phasing scheme employed the conception of overlap phases, which was inspired by
368 the TTI-4 signal strategy for conventional diamond interchanges. The phasing scheme and the
369 timing methodology had been fully tested by Hu et al. (8) for its validity under pre-time, fully
370 actuated, and actuated-coordinated control modes. The proposed phasing scheme and timing
371 methodology can also be easily modified to meet the specific DDI requirements under a variety
372 of geometric characteristics and traffic demand conditions. A case study was conducted using
373 the first DDI site in Nevada. The simulation results by VISSIM showed that the proposed
374 operation outperformed what was initially implemented by reducing vehicle delays.
375 Further research identified through this study includes the following. First, the
376 relationship between cycle length and interchange level delay needs to be further addressed.
377 Second, the relationship between traffic demands, phasing scheme, and signal timing
378 parameters, needs additional investigation. Finally, the effect of signal spacing on DDI’s
379 performance needs to also be thoroughly researched.
380
4. Bared, J., P. Edara, and R. Jagannathan. Design and Operational Performance of Double
Crossover Intersection and Diverging Diamond Interchange. Transportation Research
Record, Vol. 1912, No. 1, Jan. 2005, pp. 31–38.
5. Bared, J. G., T. Granda, and A. Zineddin. Drivers’ Evaluation of the Diverging Diamond
Interchange. Publication No.: FHWA-HRT-07-048, McLean, VA, 2009, p. 11.
8. Hu, P., and Z. Tian. Proposed Signal Control at the Moana Lane / U . S . 395 Diverging
Diamond Interchange. NDOT Sponsored Research (to be published), Reno, NV, 2012.
10. Hu, P., and Z. Tian. Advanced Signal Control Strategies and Analysis Methodologies for
Diverging Diamond Interchanges. University of Nevada, Reno. Reno, NV, 2013.
http://www.youtube.com/watch?v=VVOvzOTHUDA. Accessed Oct. 29, 2013.