| |
Home
Administration
Course Materials
Resources |
Private Notes
15 August 2007
Notes for typical class format: Include questions at the end
of each class, review of assignment or study questions at beginning of
each class.
16 August 2007
We have only two major goals here: keep cars from hitting each other
and don't make them stop at all or for too long!
Reference: Dixon's VISSIM tutorials.
19 August 2007
Comments and reflections from student assessments from Fall 2005.
1. It seemed like we spent the whole semester working on one
intersection and accomplished very little because there was no problem
in the first place.
2. Examples of what good design report should look like.
3. When using a model, make sure to help student to focus on what data
are important and relevant to the task at hand.
4. Complaints about how realistic CORSIM modeled reality.
5. General sense that communication skills are very important and
valued.
6. Be clear on assignments, show why I'm asking them to do something;
show relevance and context.
7. Discuss how team should work together: set regular meeting times,
establish means for communications, respect each others ideas, be clear
on expectations on work and meeting.
8. Be clear on use of software; step by step for new software is
important, covering both how to use, input data required, default
settings, and how to use output.
9. Labs might be too tedious; can I make them more focused and tied to
specific objectives.
10. Show examples! Provide "small" homework problems just to make
sure that students can test themselves that points are understood.
11. Consider one or more assignments or labs that students must complete
on their own.
12. Possibly schedule lab or class with Dale to show cabinet and how it
works; tie this to lecture material on this subject.
13. Consider changing team members for two projects.
14. It is important to consider background for standards: why do we set
the values that we do? Give them a chance to learn about this and
why standards are set the way they are.
15. Provide students with sufficient resources to give them support as
they learn. Reading assignments for each class and expectations
that they will be required to perform something based on these reading
assignments.
16. Consider the lecture in another room and only the lab in BEL 117.
17. [Note to myself: is the first design report really a design; do I
need to introduce something challenging to the problem to make it a
design, or is this really just learning how to evaluate an existing
intersection?]
18. Engage students more in questions and answer discussions about the
material.
19. Suggest that I pick teams based on relative experience of the
groups.
11 September 2007
Lab 4; need to correct maximum green timing process for spreadsheet
emulator. The problem appears to occur when the conflicting
detector is pressed for the second time; the maximum green timer
freezes. Also, what assumption do they make when there is no conflicting
calls.
19 September 2007
I learned another lesson in lab 5: always test the lab before doing
it. So, the problem was (and something that I should have
remembered) is the VISSIM can run at 0.1 second resolution but the NEMA
controller only uses 1.0 second resolution. This means that the
tests of zero PT and MG didn't work correctly. It also points up
the need for a set of labs that build knowledge for students of actuated
controllers. I desperately need to structure a set of building
blocks that will build students knowledge of the basic concepts.
I'm still not there yet. Ugh!
23 September 2007
Assignment 02 should be corrected to plot speed vs time, or let them
know how to collect speed profile information (using a VISSIM data
output file).
26 September 2007
The purpose of this list is to establish the body of knowledge that
I will cover in coordination during the next several weeks.
1. Motivation: Moscow CBD .avi file showing good and bad coordination.
2. Using basic queuing model, show how various arrival patterns affect
delay.
3. How do we determine arrival pattern? Robertson or T7F flow
profile model to generate downstream arrival patterns. Example
problem showing how a platoon disperses over time and distance from the
subject intersection.
4. Time-space diagram. Illustration of offsets and bandwidth.
Principles of 1-way and 2-way progression (basic relationships).
5. Coordinating timing parameters: cycle, split, and offset.
6. Phase sequencing alternatives.
7. Effects of offset on delay; examples.
8. Actuated coordinated timing parameters; 1202 Force offs;
permissive periods.
9. Problems or issues: early return to green, poor spacing, transitions,
pedestrian timing.
10. VISSIM: Fixed time example. Actuated coordinated example.
Evaluation and system performance. SILS/HILS.
27 September 2007
Following is an extended outline for coordination:
1. Motivation: Moscow CBD .avi file showing good and bad coordination.
2. Using basic queuing model, show how various arrival patterns affect
delay.
a. Standard queuing models with uniform arrivals
b. 100 percent arrivals on red and resulting delay
c. 100 percent arrivals on green and resulting delay
d. HCM arrival type model with examples
3. How do we determine arrival pattern? Robertson or T7F flow
profile model to generate downstream arrival patterns. Example
problem showing how a platoon disperses over time and distance from the
subject intersection.
4. Time-space diagram. Illustration of offsets and bandwidth.
Principles of 1-way and 2-way progression (basic relationships).
5. Coordinating timing parameters: cycle, split, and offset.
6. Phase sequencing alternatives.
7. Effects of offset on delay; examples.
8. Actuated coordinated timing parameters; 1202 Force offs;
permissive periods.
9. Problems or issues: early return to green, poor spacing, transitions,
pedestrian timing.
10. VISSIM: Fixed time example. Actuated coordinated example.
Evaluation and system performance. SILS/HILS.
21 October 2007
Notes for midterm grades:
Design report #1 - 15%
Assignments - 10%
Class discussion - 5%
Exam - 15%
Total - 45% of semester grade
22 October 2007
When introducing simulation, there are a variety of things that I
forget each time. For example, what is the value of calibration?
How many runs are needed to find a truly average value? How much
start up time is needed to allow the simulation model to reach
equilibrium? How long should a simulation period be?
And I continue to feel frustration about
either my lack of specific instructions (anticipating problems that may
come up) and their understanding of the instructions or their ability to
do the tasks as fast as I desire. Is this me? Is this them?
For example, the volume task has turned into a much bigger deal that I
anticipated.
24 October 2007
I need to give them guidance on how to use data that show optimal
values of one offset for one intersection and another value of offset
for another intersection. These could be examples from this year's
class.
28 October 2007
Notes from previous classes (CE474F02) on actuated coordinated
operation. These notes should be integrated into discussion on
actuated coordinated systems.
Class 35:
It is
worthwhile to consider several system issues that might
provide you with some useful perspectives:
- What is your
objective in providing coordinated signal timing?
Are you attempting to provide a green band for all
of the intersections in your system? Or, are there
some intersections that are "different" in nature
than the rest of the intersections, requiring such
different cycle lengths or phasing plans that they
should operate independently?
- Does
interconnection truly make sense for all of the
intersections when considering the distance between
some of the intersections in the system? If
intersections are too far apart, platoons will
disperse and there will be no structured platoon to
arrive at the next downstream intersection to take
advantage of any progression opportunities that may
exist.
- How much do you
need to balance the needs (demands) of the
individual intersections with the needs of the
system? You may encounter some trade-offs in
balancing the green splits at an individual
intersection, with the need to have more green time
on one approach for progression purposes.
- Should you
emphasize the flow in one direction at the expense
of the other direction? It may be that one
direction on the arterial is more important and thus
should receive your primary attention when
determining your offsets.
|
Class 36:
|
Yield Points,
Force Off Points, Permissive Periods
The purpose of
this class is to learn more about the three
parameters used in actuated-coordinated traffic
signal systems.
Click
here for the quiz,
with two problems on computing these three
quantities.
Let's first
consider the solution to the first problem on
the quiz. We will compute the yield point, the
force off point, and the permissive period range
in that order. The timing plan includes two
phases, one for the coordinated phases (2/6) and
one for the minor street phases (4/8).
For an actuated
coordinated system, the yield point is
the termination of the coordinated phase. It is
the only point in the cycle must always occur
during each cycle. When we review the
time-space diagram, we see that the yield point
for intersection 2 is 45 seconds after the yield
point for intersection 1. If we were dealing
with fixed time control, we would say that the
offset between the intersections is 45 seconds.
If we assume that the system zero point is equal
to the local zero point (t=0), the yield point
for intersection 2 occurs at t=45. See below
for the graphic representation of the yield
point for this intersection.
The force off
point is the point at which we must terminate
the non-coordinated phase (here, phases 4/8) so
that the coordinated phase can start at the
proper time (the beginning of the green band for
the arterial served by phases 2/6). From the
time-space diagram, we can see that the length
of the green phase is 45 seconds. This means
that the green phase must start at t=0. Since
the clearance time (yellow plus all red) for
phases 4/8 is 5 seconds, the force off point is
at t=85. See below for the graphic
representation of the force off point and the
beginning of the coordinated phase.
The permissive
period is the point in the cycle in which the
right of way can be transferred from the
coordinated phase to the minor street
(non-coordinated) phase. In CORSIM, and in some
controllers, the beginning of the permissive
period is always the same as the yield point.
Thus, in this problem, the beginning of the
permissive period is at t=45.
The end of the
permissive period is that point in the
coordinated phase that still allows enough time
for the coordinated phase clearance and minimum
green time for the non-coordinated phase, before
the force off point. For this case, the end of
the permissive period is thus 85 seconds (force
off point) minus 5 seconds (the coordinated
phase clearance time) minus 5 seconds (the
minimum green time for the non coordinated
phase), or t=75. See below for the graphic
representation of the end of the permissive
period.
See below for
all of the points represented on one graph:
29 October 2007
I need to do a clarification or addendum for
lab 8. The way that most of the students are
interpreting the lab (and I bear some
responsibility here) is to run a single offset
value for two of the intersections so that they
can't really optimize the entire system. I
need to change the instructions so that this is
much (!) clearer. I'm now going to be
working on an addendum that should get them on
the right track. The challenge is that this will
require putting off the ASC/3 lab until next
week (which is ok as we don't have all of the
problems fixed yet).
30 October 2007
So the addendum solved the problem.
This did lead to a "learning moment" for me; it
is one more piece of information that is worth
sharing with students as they proceed through
the journey: coordination requires first
and foremost a reference point from which the
offsets are measured.
There are lots of
details in VISSIM that need to be taught and
discussed. Lots of technical details:
setting up data collection points, setting up
nodes.
How to balance the
offset plots when one direction comes out better
than the other.
Is it really true
that they only need to consider two
intersections at a time? Can I confirm
this? I think its true.
31 October 2007
(notes from Class 32)
In class 32 today, the students presented
the results of lab 8, the study of the effects
of cycle length and offset on delay.
Following is a summary of my assessment of their
work.
Team 4. Splits
seem ok. Need larger fonts, hard to read
charts. How did they decide on offsets;
not clear on justification. Score: 3/5.
Team 3. Need
larger fonts. Very odd delay vs offset
curves. Good answers to questions about
these funny shaped curve. Score:
4/5.
Team 2. Very small
fonts. Ugh! Good explanations on
results. Graphs not labeled. Score:
4/5.
Team 1. Very
good presentation. Excellent graphics and
display of results. Good understanding of
the issues involved. Excellent use of
animation to identify results. Score: 5/5.
What ideas did I
get from this class:
1. Next time show examples of good and bad
presentations: fonts, charts, etc.
2. Discuss concept of travel time/offset
relationship.
3. Talk about how you can use the animation to
verify what the data tell you.
4. Why does the delay vs offset curve sometimes
show two or more humps? I would expect a
sine wave type curve that only shows one min and
one max. I should experiment with this to
see why.
5. I need to provide a path to pull all of the
calculations together from the semester into
their final report.
6. How to drive home the idea that longer cycle
lengths produce longer delays.
7. It would be good to separate these
activities: do the cycle length analysis, both
theoretically, then with simulation results, for
example. Then do splits and offsets.
More depth in each.
8. Talk about how to decide; bring in LOS ranges
or other means to determine which design value
is best. Also, talk about what is a
significant difference? How closely can
people judge differences in delay?
9. Think about what rubrics can be used to show
what is good and poor performance; this will
help make my expectations clearer.
6 November 2007
1. Some of the many details for
actuated-coordinated: Max1 times should be
larger than the split times.
2. Use detector numbering scheme as per Darcy's
suggestion for MOST.
3. Should you use detector on major
(coordinated) approach? Why or why not? |
|
|
|
|
