ML20236X231
| ML20236X231 | |
| Person / Time | |
|---|---|
| Site: | Seabrook  |
| Issue date: | 11/30/1987 |
| From: | MASSACHUSETTS, COMMONWEALTH OF |
| To: | |
| Shared Package | |
| ML20236X233 | List: |
| References | |
| CON-#487-4974 OL, NUDOCS 8712090062 | |
| Download: ML20236X231 (83) | |
Text
- _ _ _ _ - _ _ - _ _ _ _ _ _ _ _.
A lh g @lVP YY" DOCKETED USNRC
\\
'87 DEC -4 20:26 UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION OU.lC 07 EEuf.74j g)Shg['EWid;v g
Before Administrative Judges:
Ivan W.
Smith, Chairman Gustave A. Linenberger, Jr.
Dr. Jerry Harbour
)
)
In the Matter of
)
)
PUBLIC SERVICE COMPANY OF NEW
)
Docket Nos.
)
50-443-444-OL (Seabrook Station, Units 1 and 2)
)
(Off-site EP)
)
November 30, 1987
)
CORRECTED TESTIMONY OF THOMAS J.
ADLER ON BEHALF OF THE ATTORNEY GENERAL FOR THE COMMONWE ALTH OF MASSACHUSETTS ON REVISED SAPL CONTENTION NO. 31, SAPL CONTENTION NO. 37, AND TOWN OF HAMPTON REVISED CONTENTION III TO REVISION 2 (the "ETE CONTENTIONS")
Department of the Attorney General Commonwealth of Massachusetts One Ashburton Place Boston, MA 02108-1698 (617) 727-5575 8712090062 871130 PDR ADOCK 05000443 PDR
/
'1
.e A
I
[. Adler'ETE. Testimony)
' (TABLE OF CONTENTS m
Pace-I.
IDENTIFICATION OF WITNESS
.............................-l c
'A.-
' Ed u c a t i o n al' B a c k g r o u n d s....'................... '
'2
'B.
Professional Experience.......................
2 'C' Nature Of Current Work
~1.
Activities at Resource Systems Group'(RSG) 3-4 2.
.Other principals at PSG 4-5 3.
Other staff ~at RSG...........................
6 II. OVERVIEW OF THE TESTIMONY
............................. t A.
ScopeLOffWork Performed For Attorney General...........................................
6 B.
Descriptionlof How RSG Approached Work........
6-8 C.
Summary Cf The Most Important Results.
1.
.The ETEs in the NHRERP are unreliable, especially for the summer scenarios 8
2.
Many beach-goers will.be unable to get off the beach strip.(Rt. lA) for 8-10 hours 9
3.
KLD Associates significantly underes-timated actual summer weekend ETEs, which are at least 96% longer than KLD's estimate............................
10 4.
The complete set of ETEs can and should be re-done.........................
10 5.
Comparison of ETEs just to get people out of the beach areas (off Rt. lA):
RSG vs. KLD 11-13
-i
[.,
L Pace D.
Why Beach. Areas." Appear".To' Empty.
n Less : Time On - Busy Beach Days Than The' Time It:Would Actually.'Take During An Evacuation 13-14 E.-
==
Description:==
Of An Evacuation Trip From.
Hampton Beach State Park....................15-16a
' F '..
. hy.A Full ~ Beach-Area Vehicular W
Evacuation May Not Be Possible Or Even Likely................................... 16, 17 G.
.RSG Analyses _'Were Conducted Using The Same Computer Model and Inputs Used 18-19 By KLD
'III.
ANALYSIS OF THE-CONTENTIONS.
A.
KLD Had Inadequate Data On Actual.
Parking Utilization During'Truly Peak.
Periods To Accurately Infer The Area's Parking Capacity--TOH'III/ Basis A SAPL/31 Easis 20 20-21 B.
Roadway Capacities Were Mis-Estimated--
TOH III/ Bases B And C; SAPL 31/ Bases 7, 9,
15 And 18 23 1.
Traffic capacity is an important factor in determining ETEs 23
'2.
The procedure used'by KLD in pre-paring capacity estimates 23-24 3.
What RSG did to review KLD's traffic capacities 24-25
.4.
Actual achievable capacities of the evacuation. network are lower than the capacities used to calculate the NHRERP's ETEs 25 (a)
The actual road network has geometric restrictions not recognized /modeled by KLD 26-27 1
l
- ii -
- ')
R 1
-Pace
-(b)
The1I-DYNEV-r.odel appears-to apply. incorrectly the 15%'
reduction factor for-congested flow. conditions on critical.
freeway ramps...................... 27-28'
'(c)
The effects at. intersections 1of considerableLtraffic' flows' caused'by; commuters returning
'home'is not'modelede................ 28-31 (d)
The effects of disabled vehicles have?not been explicitly; con-sidered'.............................
31-34;,
(e).The effects of' inclement weather have not been adequately: rep-
. resented........................... 34-37 C.
The ETE Computations-Provided-ByrKLD Are Biased On The._ Low Side By As Much;As.25 LMinutes Because~Of.ALConcepual Error
~
In.The' Interpolation Method'They Used--
SAPL 31; TOH III 37-39 D.
The Significant Growth Projected (by.Dr'.
Luloff), For The'New' Hampshire Portion Of The'EPZ, Will Cause Increasingly Longer' ETEs Unless Appropriate Highway Improve-ments Are Put Into Place--TOH III/ Basis E and SAPL 31/ Basis 3........................ 39-41 E.
.The' Estimate Of 2.6 People Per Evacuating Vehicle, Used By KLD In Computing ETEs, Is Unrealistic--SAPL 31/ Basis'6 And'19 42-43 F.
A Number-Of UnfoundedfAssumptions About The Traffic Management Plan Were Made By KLD In Computing ETEs For the NHRERP--SAPL 31/ Basis 4; TOH III/ Bases (c) (2) and (f) 43 1.
That all traffic' Control Posts (TCPs) will be staffed immediately.............. 44-46 2.
That'all evacuees will' follow their designated evacuation routes............. 46-49 3 '.
That optimal traffic flow control will occur at intersections 49-51
[..
1
- lii -
e=
'f t
- Paae" LI V. WHEN THESE DEFICIENCIES AND UNFOUNDED ASSUMPTIONS IN THE-NHRERP's ETE STUDY ARE CORRECTED, IDYNEV GENERATES MUCH LONGER ETEs 51 A.-
Correcting Five-(5) Factors Together--
1.=
ETEs for a summer weekday withirain
-increased froml9:45 to 12:15
...............'51 2.
ETEs for a summer weekend'with good-weather i-ncreased from 6:15 to a time.
which IDYNEV 'could not' calculate' but which exceeds 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br /> 52-
?f B.
Description Of Other Sensitivity Runs...... 52, 52a f
C.
Inconsistencies.Between What KLD Modeled And What Was-Reported In The NHRERP.............. 5 2, 54 nV.
THE-NHRERP'S TRAFFIC MANAGEMENT PLAN IS UNWORKABLE..........................................
54 A.
There Is No AssuranceJThat Returning Commuters, Buses And Emergency Vehicles Will Be Able To Pass Quickly And Safely Through Access. Control And Traffic
. Control Posts--TOH III/ Basis'(c); SAPL 31/ Basis 7; And SAPL 37 54 1.
Identification checks and lane re-strictions at Access Control Posts
- will cause substantial delays and long queues for returning commuters 55-57 2.
These long queues of cars and lane restrictions will interfere signifi-cantly with the movement'of emergency buses and vehicles into the EPZ 57 3.
The placement of cones and barricades will impede all returning vehicles.....
57-58 s
4.
Ambiguity and contradictions in the descriptions of traffic control strategies in the plans may generate head-on conflicts.......................
58-59
- iv -
i, v
?
Pace
'VI.
THE NHRERP'S PROVISIONS FOR TRANSIT--
m DEPENDENT INDIVIDUALS-IS INADEQUATE --
SAPL 31/ BASES 6, 12, 13, 17; SAPL 37 60 A.
KLD Mis-AppliedLAn Adjustment? Factor FortOut-Of-Service Vehicles
'...................~.._61
.B.
'KLD's Assumpt' ion That 50% Of'Those Not' Having" Access' To A Car Will-Carpool' Is u
Not Reliab1ei................................. 62-63 r
C.
'The Use Of Either The First Market Research Survey or The June 1986 NHCDA Survey Is An Un-reliable-MethodfFor-Estimating The. Size Of The. Transit Dependent Population.......... 63-64.
D.
The Bus Mobilization Times Study Conducted By Applicant s Is U nreliable.................. 64-65 E.
The. Likelihood Of Achieving 1An Evacuation Of The Transit-Dependent In The Times-Assumed In The'NHRERP Is Remote................. 66 VII..
FINAL OBSERVATIONS A.
Why KLD's " Sensitivity Tests" Show Only Small Effects In ETEs 66 B.
The ETE Study Contained In Volume 6 Of The NHRERP Does Not Meet All The Requirements Of Appendix 4 to NUREG-0654/ FEMA-REP-1, Rev.1 68-73 C.
Many Additional Aspects Of The KLD ETE
. Study Nere Not Capable Of Our Review Because Documentation We Requested Was Either Refused Or Reported To Be Discarded 73-75 D.
This ETE Study Is Inadequate And Should Not Be Relied Upon To Protect The Public..... 75-77 l
-V
_A
b o
UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION Before Administrative Judges:
Ivan W.
Smith, Chairman Gustave A.
Linenberger, Jr.
Dr. Jerry Harbour
)
)
l In the Matter of
)
)
PUBLIC SERVICE COMPANY OF NEW
)
Docket Nos.
)
50-443-444-OL (Seabrook Station, Units 1 and 2)
)
(Off-site EP)
)
November 30, 1987
)
CORRECTED TESTIMONY OF THOMAS J. ADLER ON BEHALF OF THE ATTORNEY GENERAL FOR THE COMMONWEALTH OF MASSACHUSETTS ON REVISED SAPL CONTENTION NO. 31, SAPL CONTENTION NO. 37, AND TONN OF HAMPTON REVISED CONTENTION III TO REVISION 2 (THE "ETE CONTENTIONS")*
I.
IDENTIFICATION OF WITNESS Q.
Would you please tell us your name and current occupation?
A.
My name is Thomas J.
Adler and I currently serve as President of Resource Systems Group and as Principal-in-Charge for transportation projects.
Q.
I would like to begin this testimony by questioning you, Dr. Adler, regarding your background and the general nature of your work on this matter.
What is your educational background?
To the extent that this testimony may address matters raised in other contentions filed by SAPL, it is being sponsored by SAPL.
L
F
]
L F
1 1
l-L A.
I received a Bachelor of' Science in Civil and Environmental Engineering from Cornell Unversity in 1972, a Master of Science from Massachusetts Institute of. Technology in 1975 and a.Ph.D. from M.I.T.
in Transportation. Systems in 1976.
While completing my doctoral dissertation I was a fellow at the Joint Center for Urban Studies of M.I.T. and Harvard.
L 0
What has.been your professional experience since completing'your formal educational training?
A.
Over the period 1976 to 1986 I was a professor in the Resource Policy Center, an academic and research program of the' Thayer School of Engineering at Dartmouth College.
At Dartmouth I taught graduate and undergraduate-level courses in the areas of transportation systems analysis, transportation engineering, transportation planning, computer / mathematical modeling techniques, statistical analysis, and computer science.
I directed a program of research for-clients such as the U.S. Department of Transportation, U.S.
Department of Energy and others in the general area of computer modeling of transportation systems.
I also consulted on a part-time basis for major transportation consulting firms such as Cambridge Systematics (Cambridge, MA) and Gannett Fleming Transportation Engineers (Harrisburg, PA).
Q.
What was the nature of the research activities pursued by you and your colleagues at Dartmouth's Resource Policy Center?
A.
My central focus at the Resource Policy Center, and the focus of our academic program, was on the science of - _ -
~
,.1
- t'
. computer modeling of complex system.- WeJwere-very much L
concerned with formalizing: procedures to guide the development and application ~of large computer models.
The Center' organized-several major' conferences ~on' evaluation of' computer models of.
l energy. supply and demand,.and I personally directed reviews,
'for federal agencies, of several:large transportation models-.
'Q.
Have youLalso been' involved in the actual. development-and application of transportation-related models?
A.
Yes.-LFor example, at Dartmouth I directed a project for the U.S.
Department of Transportation whose product.was.a.
simulation model of transportation energy use.
'I also direct.ed another project which. involved development.of a traffic network
. simulation model.
I have also been heavily involved in the development of techniques for collecting data to be used as input to transportation models.
I was co-author of a manual-distributed by the Urban Mass Transportation Administration which describes.new techniques for using behavioral intentions
{
data;in transportation models.
Q.
For how long have you served in your current position 1
as President of Resource Systems Group?
i A.
I have been president since October 1986, when Resource Systems Group was formally incorporated.
O.
What is the nature of your current work at Resource i
L Systems Group?
A.
Besides my administrative responsibilities, I have maintained research and project activities similar to those that I pursued while a professor at.Dartmouth.
For example, I i
j _ _ _ _ _ _ _ _ _ _ _ _ _ - - - - _
e l
l
>vh l
l am under contract with the Ohio Department of Transportation to 1
independently review a large transportation network modeling
~
f project.'I am also directing a project to estimate input parameters for an urban transportation model for Tampa, j
i Florida,.under subcontract to the transportation planning firm of Parsons Brinckerhoff Quade1& Douglas.
I have been involved with one of our other principals in review and enhancements of
'an unclassified model developed for use by the office of the Joint Chiefs ofJStaff in the Department of Defense.
And I have also worked for numerous municipalities and private concerns in 1
projects involving traffic impact analysis.
I have served as Principal-in-Charge at Resource Systems Group for the project I will be describing in this testimony, our Seabrook Evacuation Time Estimation project.
Q.
Besides you, who are the other principals at Resource i
1 Systems Group and what are their backgrounds?
A.
Dr. Dennis Meadows is one of the other two founders and principals of Resource Systems Group.
Dr. Meadows is also Professor of Engineering and Director of the Resource Policy l
1 l
Center at Dartmouth College.
He is an internationally recognized expert in the field of computer modeling and simulation, and he is the author of several books and numerous journal articles documenting and evaluating applications of computer modeling of complex systems.
He has twice served as l
Senior Scientist at the International Institute for Applied
)
Systems Analysis in Laxenbourg, Austria, most recently as Program Director.
Dr. Meadows also serves as executive l
_4-1
_-_________J
h director of INRIC (International Network of Resource Information Centers), a confederation of over twenty international institutes involved in computer modeling and l
resource analysis.
prior to joining the faculty at Dartmouth he'Was a professor at M.I.T.
His activities at Resource Systems Group have' included development of simulation models for management training, and he has served as an advisor on our.
Seabrook Evacuation Time Estimation project.
Dr. Colin High, the other principal of Resource Systems Group, is a professor in the Dartmouth Environmental Studies program.
He has formal academic training and extensive experience in the fields of geography and environmental science.
He.also has directed several major projects which involved the development and application of computer simulation models.
At Resource Systems Group, Dr. High has directed resource modeling projects for international clients and serves as an advisor on projects involving remote sensing.
He has taught courses in interpretation of aerial photography and directed a NASA-sponsored project in that field.
He was directly involved in the interpretation of aerial photographs of the Seabrook area seacoast which were compiled for the Seabrook Evacuation Time Estimation project.
More detailed resumes for all three of the Resource Systems Group principals are attached to this testimony (Attachment 1).
Q.
Are there other professiorhl staff at Resource Systems Group who have contributed to the work on your Seabrook Evacuation Time Estimation project?
_ 5 _.
p -
A.
Yes.
We currently have seven full-time professional-L staff' members-who work on modeling and transportation-related projects.
Most of these individuals were involved in some part of the Seabrook Evacuation Time Estimation project.
Stephen Hastings, a Senior Project Associate with a Master of Science degree from Dartmouth in System Simulation, conducted most of the computer analyses with the assistance of Jeffrey Sidell, who also has his degree from Dartmouth in Computer Science.
Two other Project Associates, Leslie Rimmer and Megan Haney, assisted in several data analysis tasks.
All of these tasks were performed under my direct supervision.
During my testimony I will refer to these Resource Systems Group staff and principals as "we".
The work is familiar to me in that, while I did not perform all of the work personally, I did personally specify and supervise that work.
II.
OVERVIEW OF THE TESTIMONY 0
Would you please describe briefly the scope of the work that Resource Systems Group performed for the Massachusetts Department of the Attorney General?
A.
We were asked generally to review and evaluate the validity and reliability of the Seabrook Station Evacuation Time Estimates and the adequacy of the Traffic Management Plan Update as incorporated into NHRERP Volume 6.
Q.
How did Resource Systems Group approach this work?
A.
We divided the work into three general tasks:
1)
Evaluate the accuracy / reliability of the assumptions and data that were used as inputs to the analysis reported in volume 6; )
l
^:
n.
.F
'2)
Determine'the effects.on Evacuation Time Estimates (ETEs) of! those inputs which we found to be-inaccurate or otherwise
~
unreliableLand, where possible,. insert
- the better assumption /more accurate-data; and
- 3). Assess the importance of factors that were not. explicitly considered in th'el Volume 6 ETEs and. determine their effect on the ETEs.
n In completing parts of the first task we relied in part on data and analyses performed,by other consultants retained by the Massachusetts Attorney General.
Specific-contributions will be detailed later in this testimony.
In several very important areas, it appeared that'the assumptions nnd inputs used in Volume 6 were not appropriate and so I wantedito test the effects of alternative. assumptions / inputs on the ETEs.
In order to allow direct comparisons of these
.results with the ones reported in Volume 6, the best procedure would be to employ the identical computer modeling protocol used by KLD in preparing those ETE estimates.
It was also important to independently verify the KLD results, as is commonly done in other computer modeling applications.
- Thus, we concluded that we needed to have direct access to KLD's computer model.
On July 8, 1987, KLD provided us with a compiled version of the computer program and computer tapes of Ethe input files'it used for the ETE study documented 'in Volume 6.
Over the intervening period we have completed over 75 separate model runs and sensitivity tests using I-DYNEV, the computer model developed by KLD and used by KLD in preparing ETEs for the NHRERP, Revision 2.
While the compiled I-DYNEV t
l l.
program allowed us to complete several simulation runs, it did not allow us to verify the implementation of the methods described in Volume 6 as we would have been able to do had we been provided the computer source code.
Although we had I
requested I-DYNEV's source code, KLD refused this request.
On reviewing Volume 6, I also' identified several important factors that were not explicitly considered in the KLD analyses, and we performed additional analyses, to evaluate and quantify the effects of those factors.
This testimony will detail the work that we completed and the results that were obtained which are germane to this proceeding.
Q.
In the work that you did, did you apply methods and procedures which are generally accepted in you profession?
A.
Yes.
Q.
Before you go into further detail describing your work, would you please first summarize the most important results?
A.
There are several findings which individually are quite important, but the most important single conclusion deals
)
with the overall reliability of the ETEs presented in Volume 4
6.
In our view those ETEs are unreliable, especially for the summer scenarios.
The results of our study indicate that the overall time to evacuate the entire EpZ on a summer weekend when the beaches are crowded is not 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> and 15 minutes as l
KLD has estimated, but is at least 11 hours1.273148e-4 days <br />0.00306 hours <br />1.818783e-5 weeks <br />4.1855e-6 months <br /> and 15 minutes.
We l
i also found that the vehicles in the beach areas are the ones u_-________-____-.
l which experience the longest.ETEs and that, because of the extremely limited number of roads leading off the beaches, cars will back up in such long queues there that many beach-goers l
will be unable to get off the beach strip (Rt.1A) for 8 to 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br />,' depending on where they parked their cars.
The principal author of the KLD study has asserted that the accuracy of the ETEs for a summer weekend, as reported in Volume 6 of the t3HRERP, is plus or minus 10 percent.
In fact, we found, using the same computer model employed by Volume 6's authors, KLD Associates, that the Volume 6 ETEs consistently and significantly underestimate actual ETEs under both plausible and likely evacuation conditions.
One reason why KLD has seriously underestimated ETE's is that it made a series of assumptions about aspects of any evacuation which are unlikely to prevail in real-life conditions.
Two such assumptions are (1) tnat all evacuating vehicles will take the routes out of the EPZ which the plans prefer; and (2) that all traffic control posts ("TCPs") will be staffed, with barriers and traffic cones in place, at the moment an evacuation begins.
But even if the summer weekend evacuation were to proceed exactly as planned, which we do not believe would happen, we found that ETEs will still be significantly longer because some of the critical inputs appear to be in error.
In particular, as the testimony I presented jointly with Drs. Befort and High has shown, the number of vehicles which can occupy the beach areas on peak summer weekends is at least 50% greater than reported in volume 6. _ _ _ _ _ _ _ _ _ _ _ -
l l-l Thus, KLD's estimates of evacuation times for its summer weekend scenarios (i.e., Scenarios 1 and 2) reflect not the times'to evacuate beaches which are at 100% of capacity, but the times to evacuate beaches which are only about 2/3 full.
In addition, the KLD analysis ignores altogether (i.e., does not even factor into the model) certain important real-life elements such as the additional traffic created by thousands of drivers who will be returning home to pick up other family members.
Together, the effect of just the few factors I have specifically mentioned above is to increase the ETE by more than 96% over KLD's estimate--from 6:15 to 12:15--for the summer weekend evacuation of the entire EPZ.
Note that I say
.more than.
In doing our model runs, some of the effects that lengthen evacuation times were not modeled for all locations in the EPZ.
For example, we evaluated the effects of returning commuters only at two locations, representing only a small fraction of the total of such vehicles.
Had we modeled this effect for all locations, ETEs would have been shown to be even longer than reported here.
We did not do this additional modeling because the objective of our work was to determine the overall accuracy and reliability of the Volume 6 ETEs, not to prepare a full set of " alternative ETEs."
As I describe later in my testimony, my expert opinion is that the complete set of ETEs for Seabrook Station's EPZ can and should be re-done in a manner that produces a much more reliable set of ETEs than those in Volume 6. _ _ _ _ _ _ - _ _ _ _ _ _ _ _ -
While most of: our analyses assumed full evacuation of the EPZ (Region 1), we did evaluate the effect on ETEs for a beach-area-only evacuation (Region 10) for the summer weekend scenario (Scenario 1), with the effects described above included.
In this analysis we also assumed a 50% " shadow" evacuation of the remainder of Region 1, following the recommendation about this factor described in the testimony of Drs. Zeigler, Johnson, and Cole.
The results are shown in Figure 1 (next.page) in comparison to the values shown in Volume 6, Table 10-9 (p. 10-11).
The times shown, in all cases, represent the times to evacuate the beach areas only, i.e.,
to get all cars off of Rt.lA or the respective beach roads.
As this figure shows, we estimated beach evacuation times approximately twice as long as reported in Volume 6.
The time it takes to evacuate to 2, 5, or 10 miles away from the beaches is somewhat longer.
We also found several important areas where omissions and ambiguities in Transportation Management Plan elements of Volume 6 will result in an overall reduction in the efficiency j
l of the evacuation, compared to what would be q
i I
possible/ desirable.
For example, the management of emergency and commuters' vehicles returning into the EPZ has not been carefully considered, with the result of likely extended delays for all of these vehicles at several Access Control Points (ACPs) and Traffic Control Points (TCPs).
1 Q.
When you mention the terms " Scenario" and " Region" j
what are you referring to?
l l
l,
\\
A.
In Volume 6,~ETEs were calculated for different conditions depending on season, day, time of day and weather.
A given combination of these factors constitutes a " Scenario".
A table from Volume 6 listing " Scenarios" used is attached to this testimony (Attachment 11).
" Regions" are those subparts of the EPZ for which ETEs were calculated for each " Scenario".
A table listing the " Regions" used in Volume 6 is attached to this testimony (Attachment 12).
Q.
Figure 1 is titled, " Time Required to Evacuate Beach Areas."
Would you explain further what this figure shows?
A.
Figure 1 shows the amount of time required to clear the immediate beach area only under two evacuation conditions:
beach closing (upper part'of figure) and a general evacuation of the entire Emergency Planning Zone (lower part of figure).
The amount of time required for vehicles to clear out of the beach area is longer in the general evacuation condition because more cars enter the network in front of the evacuating beach traffic.
For each evacuation condition, two times are presented for each of the beach areas:
the results reported in Volume 6, and the ones that we computed with the assumptions that I described above.
Note that all of the times shown represent the amount of time required to clear only the roads immediately adjacent to the beach.
Fci Hampton, Seabrook and i
Salisbury, they generally represent the amount of time it would take to clear all evacuating vehicles off Rt.lA.
There are two other important points of clarification regarding these time estimates.
First, all 'of the Volume 6._-_______________a
_ - _ - _ - - _ = _ _---______ _____ _____ -
evacuation' time estimates contain a. systematic bias due to an inappropriate interpolation. method that underrepresents evacuation times by up to 30 minutes (this will be discussed later in this testimony).
The "New Estimates" reported in Figure 1 are calculated using a method'that eliminates this bias.
Second, all the times presented in Figure 1 (and in Volume 6) actually represent the time that would elapse after an assumed general evacuation message that, itself, is assumed to occur 25 minutes after the beach closing message.
So, for example, the 5:40 reported in Volume 6 as the evacuation time for the Hampton beach area actually represents 6:05 after the beach closing message.
The use of this general evacuation order as the base time for comparison is, however, misleading when applied to the beach closing-only condition since a general evacuation order is never issued.
We have reported in Figure 1 beach closing evacuation times as they were calculated in Volume 6.
So, to determine the amount of time that beach evacuees will actually be "on the road," 25 minutes should be added to all of the times shown in Figure 1 (and in Volume 6).
Thus, for example, if only the beaches areas are ordered to be cleared, we found that it will take 8:40 to do so for Hampton Beach and 5:40 for Seabrook Beach on days when the beaches are
. full.
Q.
You have shown beach evacuation times of over B hours for Hampton's beach area and 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> for Salisbury's; yet, in l
a cense these beaches " evacuate" even on the busiest days in much less time than this.
How do you explain the difference? J l
r - -
A.
There are several important differences between the emptying of the beach area on a busy summer day and the circumstances of an emergency evacuation.
First, as reported in the testimony of Dr. Albert Luloff, data from a recent beach survey by AEL Associates show that on a " normal" day vehicles leave the beach area at a relatively uniform rate over a 5 hour5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> period from 1:00 PM to 6:00 PM.
However, in an emergency evacuation, everyone would be attempting to leave the area almost simultaneously.
Second,.over 20% of the individuals on the beach have come by a means other than driving a car (most by walking) and presumably will return to their accommodations also without driving a car.
Almost 30% of the beach visitors are staying in the area more than 3 nights.
Neither of these populations would have cars on the road during a " normal" afternoon period when the beach empties.
By contrast, all l
would be part of the traffic stream in the event of an emergency evacuation.
In addition, all year-round residents and area workers would be leaving at the same time.
- Finally, there will be additional individuals in the areas between the 1
l beaches and the major north-south highways who will be evacuating and who would normally not be on the highways at that time.
In summary, the major difference between the emptying of the beach area on a busy summer day and the conditions of an emergency evacuation is a substantial increase in the number of vehicles leaving the area and a substantial reduction in the time interval over which they will try to enter the road network... _
i Q.
Could you describe how an evacuation trip from the Hampton Beach State Park would likely' proceed in the general evacuation scenario on a. busy summmer day?
A.
Assuming everything proceeded according to the Volume 6 Plan for a Region I/ Scenario I evacuation, a beach closing message would first be issued and all beach visitors would be
' instructed to leave:the beach area.
Those who were able to reach their cars the most quickly would likely be the first to exit the lot.
If they were to follow their designated evacuation route, they would turn right (north) onto Rt.lA and continue to Rt.51 West.
For the first few vehicles, this would be a relatively quick journey.
However, those who had to "round~up" a family of several children would be blocked from exiting the parking lot by a substantial queue of the 1,500 other vehicles attempting to exit this lot onto Rt.1A.
Others who parked at locations to the north of the State Park-lot would also create a substantial' queue to the north on Rt.1A.
Only 25 minutes after the beach closing, a general evacuation would be ordered and additional cars would fill the roadways to the' north and west of the State Park.
According to the data provided in Volume 6, it would take 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> and 5 minutes (from the time of the beach closing message) for the last cars out of the Hampton Beach State Park lot to make the 1 and 1/2 mile journey north to Rt.51.
This translates to an average forward progress of approximately 1 and 1/2 car lengths per minute.
According to the new estimate described above, this same trip would take 9 hours1.041667e-4 days <br />0.0025 hours <br />1.488095e-5 weeks <br />3.4245e-6 months <br /> and 50 minutes, representing forward - - _ _ _ _ - _ _ _ _ _ _ _ -
=
progress of less than one car length per minute.
In fact, progress.would not be continuous, as some vehicles would wait for long periods of time with no movement at all, and then move ahead in spurts of several car lengths.
Figure 2 (next page) shows the number of vehicles which would remain along Rt.lA in Hampton over the period following a beach closing announcement (and subsequent general evacuation).
According to the I-DYNEV runs which are reported in Volume 6, over 3,000 vehicles (over 7,000 people) will still be stuck along Rt.lA in Hampton's beach area 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> after the beach closing.
According to our new estimate, approximately that same number would remain after 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> and over 18,000 people would be in the queue along Rt.1A after 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />.
The length of the delays and the extent of congestion would represent travel conditions well outside the realm of any of the evacuees' prior experience.
Data from the AEL Associates survey, reported in the testimony of Dr. Albert Luloff, indicate that substantial numbers of evacuees will simply abandon their cars if little forwar6 progress is made over an extended period of time.
All of the I-DYNEV runs reported in this testimony assume that everyone in the evacuation zone is j
able to finally evacuate.
However, there are inherent behavioral uncertainties in how evacuees collectively and individually will respond to these conditions.
I am unable to state that full evacuation is likely or even possible without substantial intervention beyond what is described in volume 6, under the conditions that would exist. J
i j
Q-Would you explain your last statement further?
q A.
Yes, certainly.
Given evacuation speeds of less than t
one car length per minute for periods of time up to 8 or 9 I
hours (just to get out of the Hampton Beach State Park parking.
lot, travel.up Rt.1A for a mile and a half, and get on Rt.51 West), I am uncertain that a vehicular evacuation will work.
I just do not know whether people will stay with their cars that long, traveling much slower than most of them can walk.
The evidence I have, from the beach survey by AEL Associates, indicates that many will indeed' abandon their cars.
The problem, of course, is that not everyone can walk out.
There are many beach-goers with infants and small children, as well as many who are elderly, infirm, handicapped or simply not fit enough to walk a few miles.
Will those who abandon their cars leave their cars in positions which birck those who remain with their cars?
I simply do not know.
But it is such a real possibility, particularly for a place like the Hampton Beach St. ate Park lot, from which the nuclear plant is clearly in view, that prudent planners need to plan additional layers of assistance to address this circumstance.
I am particularly concerned about the State Park lot for the following reason:
many of the cars in that lot are at the end of the evacuation I
queue for the entire south beach area in Hampton, those cars will sit there in the lot, not moving at all, for hours before the-rest of Hampton's beach area clears out enough to allow the cars in that lot to begin emptying onto Rt.1A.
During the l
hours before the lot begins to empty, many cars may well be I
I _ _ _ _ _ _ _ _ _ _ _
i
L r
abandoned, and this has the real potential of creating l
" gridlock";insi'de the lot unless there is substantial additional planned intervention.
The NHRERP Traffic Management Plan calls for only a single traffic guide to be stationed at
.the point where the Hampton Beach State Park lot empties onto Rt.lA, and that guide's job is described as follows:
'"[e]ncourage all traffic.to move north along Route 1A."
(See Vol.
6, p.'I-1).
More'than a single emergency worker certainly needs to be stationed at the Hampton Beach State Park to deal with the potential for " gridlock" there, and emergency strategies need-to'be devised to address this egress problem, as well as to address the needs of hundreds, if not thousands, of beach-goers who may abandon their cars and attempt to walk out.
Q.
You stated that your analyses are based on the same computer model used in the preparation of Volume 6.
Would you explain this statement further?
A.
A major issue in this~ case is the reliability of the numbers and assumptions that serve as inputs to the ETE calculation.
I was retained by the Massachusetts Attorney General's office to evaluate the effects of these numbers and assumptions without changing the basic modeling methodology employed in Volume 6.
The reliability of the computer model used to calculate the ETEs, the I-DYNEV system, is a separate issue that is addressed in some detail by Dr. Ceder in his testimony but which I did not pursue in detail for the purposes of my work.
I will make some general comments on I-DYNEV later 1
in my' testimony, but the major focus of my work.was on testing the effects of alternative inputs to the model rather than on critiquing the model itself.
To be more specific, we obtained from KLD a copy of the I-DYNEV model system which was verified by KLD staff to be functionally identical to the version that they used in preparing Volume 6.
We also obtained a full set of the inputs used by KLD for all of the results reported in Volume 6 as'well as inputs used in KLD's further sensitivity tests.
In all cases that I will describe here, we used this version of I-DYNEV without modification and began with the exact inputs used for Volume 6, modifying these inputs only as appropriate for each individual computer run.
III.
ANALYSIS OF THE CONTENTIONS Q.
Have you read the various contentions filed by the interveners regarding ETE issues, specifically Revised SAPL 74 Contention 31, Town of Hampton Revised Contention III and SAPL 37?
A.
Yes.
Copies of these contentions are attached to this testimony (Attachment 2).
Q.
Would you please detail the findings of your work as they relate to these contentions?
A.
Certainly.
Since some of the important issues are raised in more than one of the contentions that you listed, I will organize my discussion around issue areas and cite the relevant contentions / bases as I proceed.
I will start with a _ - _ _ _ _ _ - _ _ _ _ _ _ _ _ - - _ _ - _ _ _ _
central issue With respect to the ETES, the KLD estimate of beach (" transient") population'in the Seabrook seacoast area.
For the purpose of computing ETEs, KLD was'primarily interested in computing the vehicle population in the beach areas.
To do this, they relied on several series of aerial photographs from which the beach area's " parking capacity" was estimated.
According to Volume 6, pp. E-4 to E-5, this " capacity" includes the stalls in all marked parking lots, curb space along roads, other.open accessible lots, private driveways, front yards and back yards., KLD's interpretation of the resulting estimate is given on p. E-5 of Volume 6:
We believe that these capacity estimates represent a reasonable upper bound to the number of possible parked vehicles in the indicated areas.
Town of Hampton TOH III/ Basis A and SAPL 31/ Basis 20 contend that KLD had inadequate data on actual parking utilization during truly peak periods to accurately infer the area's " parking capacity."
To assess this contention, the interveners first obtained the photographs used in KLD's analyses and requested additional detail on KLD's methods so that the Volume 6 estimates could be replicated.
Unfortunately, KLD discarded the detailed worksheets on which they recorded their counts
)
from each slide.
Having these worksheets would have allowed us
)
I to conduct a more complete audit of their work.
While actual l
i car counts from the KLD August 1985 photos were replicated approximately (as described in the testimony of William i
Befort), the " parking capacity" estimate could not be
i replicated from the protocols recounted in Mr. Lieberman's July
?
2, 1987, deposition.
Two aerial overflights of the Seabrook.EPZ beach' areas were conducted by William Befort in July 1987 to provide additional photographic data-for our work.
Interpretation of that photographic material gave two important sets of data points for our ETE work:
an estimate of the actual number of vehicles present in the beach areas on July 5, 1987, and an estimate of " parking capacity" as defined in Volume 6.
We used the latter. estimate, in a way consistent with the methodology described in Volume 6, to recalculate (using I-DYNEV) ETEs for the summer weekend evacuation scenario.
For this I-DYNEV run, we used all of KLD's inputs except those which reflected the beach area's vehicle capacity, where we substituted-our new data.
The results are shown in Figure 4 (see the bar labeled
" Beach Parking Capacity").
Even though the total number of evacuating trips for the entire EPZ is only 15% higher than the estimate used.in Volume 6, the ETE increases by 51% (from 6:15 to 9:25 for Scenario 1 full EPZ) just by changing this single input factor, because the beach population disproportionately loads the most difficult bottlenecks in the roadway network.
Q.
Who actually prepared the vehicle population and
" parking capacity" estimates from the July 1987 aerial photographs?
A.
Drs. Colin High and William Befort were directly responsible for that work, though I was involved collaboratively in developing the protocols used and in i l
'l reviewing the work, There F s been separate testimony filed which outlines the proce?ur n, and results of this work.
Q..
Dr. Adler, befott move on to other issues contained s
in the contentions, were there other data and analyses prepared by other consultants to the Massachusetts Attorney General which you used in your work?
A.
Yes. ' Social Data Analysts conducted a telephone survey of EPZ residents which provided demographic data and information on likely evacuation behavior.
This. survey responds directly to the concern raised in SAPL 31/ Basis 6, which asserts that the survey relied upon in the KLD study has a large non-response bias.
The Social Data Analysts survey also collected important information that goes beyond the scope of'the survey relied upon by KLD.
Dr. Stephen Cole has submitted testimony detailing the methodology employed in this survey.
Resource Systems Group was provided with tabulations of the survey responses, a data tape containing the individual computer-coded responses, and the original written survey forms.
AEL Associates prepared resident population and employment estimates and projections for the New Hampshire EPZ towns.
They also conducted a beach survey which provides demographic information and indications of likely evacuation behavior.
Dr. Albert Luloff's testimony details these data and accompanying analyses.
I also worked closely with Dr. Avishai Ceder, who has been reviewing some of the traffic capacity, traffic flow, and engineering aspects of Volume 6 and of the I-DYNEV nodeling system. - _ _ _
Q.
Did you do any work related to the contentions dealing with the effects of roadway capacity on ETEs?
A.
Yes.
Both TOH III/ Bases B and C and SAPL 31/ Bases 7, 9,
15 and 18 deal with the issue of how capacities for the area's roadways were estimated in volume 6.
In fact, those bases cover many different issues related to roadway capacity which we evaluated.
Q.
Is traffic capacity an important factor in determining ETEs?
A.
Yes, definitely.
Generally, the evacuation time is a direct function of the capacity of the constraining roadway sections.
A simple example illustrates this point.
It would take 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> for 10,000 vehicles to evacuate through a roadway section which has a capacity of 1,000 vehicles / hour (10,000 l
divided by 1,000 = 10).
If the capacity of this roadway section were 25% lower (750 vehicles / hour), it would take 13 and 1/3 hours to evacuate the same 10,000 vehicles (10,000 divided by 750 = 13.33), or 33% longer.
So, roadway capacity has a direct effect on ETEs.
Q.
Would you summarize the description contained in l
Volume 6 of the procedure used by KLD in preparing the capacity estimates in Volume 6?
A.
As described in Volume 6, p.
1-10, KLD conducted a survey of the " entire highway system within the EPZ and for some distance outside."
KLD claims this survey detailed the numerous characteristics of each highway section which are necessary to estimate traffic flow capacity.
KLD then ___ _ - _ _ _ _ _ _ - _ _ _ _ _
estimated maximum flow capacities for each of four general roadway types and, separately, for freeways and freeway ramps, based on procedures outlined in the Highway Capacity Manual (1985).
'In addition, "such estimates were obtained empirically at representative intersections throughout the EPZ" (p. 3-3).
These capacities are decreased by I-DYNEV by 15% "when the traffic is moving under congested conditions" (p. 3-9) to reflect the detrimental effects of driver behavior under those circumstances.
KLD also says that capacities at intersections were adjusted to' reflect the' effects of evacuating traffic flow (if any) from competing approaches.
The capacities were further reduced in scenarios which assume inclement weather:
by 20% for rain and by 25% for snow.
Q.
What did you do to verify and review the work related to traffic capacities performed by KLD?
A.
I used a three-step review process.
First, I wanted to verify the survey data that KLD had collected on the area's roadway characteristics and KLD's empirical observations of traffic flow.
However, KLD had' discarded all original records of their roadway survey and had not maintained formal records of their traffic flow observations.
Rather than replicate the full survey, I personally visited the area and inspected several key intersections to determine whether they were accurately represented in the Traffic Control Post (TCP) and i
Access Control Post (ACP) diagrams (see Vol.
6, Appendices I
]
i and L) and whether they were properly represented in the i
I I-DYNEV model.
On one occasion Dr. Ceder and I toured the area j
i 1 _ _ _ _ _ _ _ _ -
together. ;As I will describe below, we found some important discrepancies.
The second step of the review process involved checking I-DYNEV-to verify that the capacities specified were properly reflected.in the model's outputs.
We again found discrepancies which I will detail.
Finally, I worked with Dr. Ceder to determine whether the capacities assumed by KLD are, as they assert,
" conservative estimates" (Volume 6, p.
3-3, emphasis in original), and, generally, whether those capacities are reasonable for the conditions that they are intended to represent.
Dr. Ceder will describe his work on this issue in his testimony.
Q.
Would you now please detail your findings related to roadway capacity estimates?
A.
We found overall that the actual achievable capacities of the evacuation roadway network are lower than.the capacities represented in Volume 6, for five primary reasons:
(1)
The actual road network has geometric restrictions not recognized /modeled by KLD; (2)
The I-DYNEV model appears to apply incorrectly the 15% reduction factor for congested flow conditions on critical
{
freeway on-ramps; j
l (3)
The effects at intersections of considerable traffic flows caused by commuters returning home is not modeled; (4)
The effects of disabled vehicles have not been explicitly considered; and p
I (5)
The effects of inclement weather have not been adequately represented.
4 l
1 l L
('<
1.,
?
I i
The first-listed problem is best illustrated'by l
reference to one of the critical bottlenecks'that' appears.in the summer weekend evacuation scenario, the intersection Rt.110-and I-95.
This intersection is. critical to the beach area evacuation because Rt.110 serves simultaneously as:
(1) the primary evacuation route for a large number of beach evacuees.
from Salisbury, (2) a planned " overflow" route for Seabrook beach (see p.
1-31), and (3) the. primary return route for commuters-who work in areas southwest of the EPZ and who are returning to both New Hampshire and Massachusetts towns within the EPZ.
The Traffic Control Post Diagram (Vol. 6, p. I-19)
(Attachment 3 hereto) for this intersection fails to show in the southwest quadrant a wide, grassy median with raised 6 inch I
curbs that connects the two triangular islands shown.
A i
l modified version of this TCP, which shows this impediment, is 1
)
attached to this testimony (Attachment 4).
As routed, in order to access I-95 southbound from the left turn off Rt.110 westbound, evacuating vehicles would have to travel over this median.
In fact, these vehicles are being routed onto an of f-ramp which is posted wit h large "Do Not Enter" and " Wrong
(
l Way" signs.
Vehicles exiting I-95 along this ramp are being
]
only " discouraged" (not prohibited) from making this movement, which may result in head-on collisions.
The I-DYNEV model assumes that the capacity of this wrong-way use of an off-ramp, involving trave: sal of a raised, grassy median is the same as for a normal, paved freeway on-ramp.
This assumption grossly - - _
__--______-___Y
p I
overestimates the achievable capacity under these conditions.
It is a reasonable assumption that a passenger car will slow to 5 mph to climb the curb and that approximately a 25 foot gap will be left between cars to allow for lurching after the curb is negotiated.
In these circumstances, the capacity will be lowered to 660 vehicles / hour, almost half (56%) of the capacity assumed ~by KLD.
A run of I-DYNEV was made in which the capacity of this ramp was reduced to 660 vehicles / hour.
All other KLD inputs remained unchanged.
The result was an evacuation time of 8:50, a 41% increase in the overall ETE.
(Region 1, Scenario 1).
Importantly, the above analysis does not consider the fact that there are many camper-trailers (over 400 in one Salisbury lot), low-clearance sports cars, and other vehicles that simply may not be able to traverse the median.
Undoubtedly, some will try to do so, and these vehicles would further impede traffic flow.
Nor does the Volume 6 analysis consider the effects of rainfall or snow storage on the ability of vehicles to cross the median.
Q.
The second capacity-related problem that you identified deals with the I-DYNEV modeling of congested flow on ramps.
Would you please describe the work you did related to this problem?
A.
Yes.
Volume 6, p.
3-11, says that freeway ramps were assumed to have a capacity of 1170 vehicles / hour and that the capacity is reduced to 85% of this value under congested flow -
conditions _to account for congestion effects.
The resulting capacity should be 994 vehicles / hour.
We were provided by KLD with their actual computer output of the run which represents the summer weekend evacuation (Scenario 1).
This output contains many. details not reported'in Volume 6.
In checking through this output, I. discovered that the actual simulated discharge rate through I-95 ramps which serve as bottlenecks to the evacuation ranged up to 1096 vehicles / hour during periods of congested (Service Level F) flow.
This represents 10%
higher capacity than what is stated was used in Volume 6.
Since this overstatement of capacity occurs at a network bottleneck, the result is a proportional understatement of ETEs.
Q.
The third capacity-related problem that you listed earlier deals with the effects of commuters returning home.
Would you please describe the work that you did related to this problem?
A.
The KLD analysis, as reported in Volume 6, includes directly only the trips of commuters who work in the EPZ but live outside.
That analysis considers the effects of other i
commuters returning home only in a very limited way--by assuming that the capacity of two-way road sections will be reduced to a level corresponding to a 90%-10% split of evacuating vs. returning traffic.
The trips of commuters who both live and work within the EPZ and who must travel along with evacuating traffic are not considered at all.
- However,
)
the most significant effects of the returning traffic are (1) at intersections, where returning traffic streams will in many l
1 i l
l i
'l
. cases directly conflict with evacuating traffic, with resulting significant declines in capacity for evacuating flows and (2) along evacuation routes where commuters' vehicles are attempting to travel in the same direction as; evacuating traffic.
Our analysis proceeded first by! evaluating the general magnitude of this problem, then developing case examples to illustrate the effects on ETEs.
Volume 6 states (at p.
4-9) that 95% of the commuting population would leave work within 30 minutes of the evacuation order.
This would create an immediate surge.of traffic that would be far greater than any experienced under " normal circumstances."
Data from the Social Data Analysts survey shows that, under normal conditions, only one-third of the EPZ's resident workers leave work in the afternoon peak one-hour period (4:00 to 5:00 p.m.).-
The remainder of the EPZ's, resident workers leave work during other times of day, generally between 3 and 6 p.m.
See Figure 3 (next page).
Thus, the " normal" rush hour reflects only a small percentage of the commuter traffic activity that would occur under the weekday evacuation scenario.
Therefore, contrary to KLD's assumption, in a real evacuation it would take considerably longer than normal for the commuting population to return home.
Some of the adverse effects of " return home from work" trips on the evacuating traffic flow were represented in the Volume 6 study by using an assumption of 10% counterblow traffic in the calculation of roadway capacities, as described 4
t t
in Volume 6 at p.
3-8.
However, these " return home from work"'
tripsswere not._ explicitly,modeled in any offthe evacuation scenarios,:nor-werenthey modeled,as affecting' intersection.
capacity, even.though'.'they~would addfa significant. number-of tripsitoftheiroad network during an' evacuation.
Further, the assumption ~of only'10%' counterblow traffic does not' appear to be realistic because it" reflects a serious undercount of; return
\\
- trips.:. Table 5-3 of Volume 6 (p. 5-7) indicates that.1986 employment'within the EPZ was 68,084 and that 31,298 of those
- employees reside outside the EPZ.
Since all' evacuating trips by permanent residents are modeled as originating from<the
-place of permanent residence, KLD assumed that many of these-36,786 commuters return home prior to evacuating.
Using the average vehicle occupancy of 1.16 for employees, as provided.in Table'5-3 of Volume'6 (p. 5-7), a total number of 31,712 trips home-could be generated by people returning home from' work within the EPZ.-
Information in Section 5 ofLVolume 6 indicates that 26,420 EPZ residents commute to workplaces outside the EPZ.
l Using the 1.16 employees'per vehicle figure mentioned above,.
the potential exists for 22,776 trips from places of work outside the EPZ to places of residence within the EPZ.
- Thus, t
- in total, up to 54,488 commuter trips home could occur that have not been explicitly modeled in any of the Volume 6 evacuation scenarios.
These trips would generate additional crossflow, counterblow, and evacuation flow.
In the cases of Scenarios 3 and 4, mid-week scenarios when employment is at L._
j t
l w
100%,.54,488 commuter-tribs home would represent:a153%. increase in.the total 1 number.of trips lthat KLD modeled-as occurring'
'during~an evacuation.
Most of these commuter-trips woul'd.begin.
.over.a short period ofstime'at:the onset'of,the evacuation, causing a heavy flow of" commuter traffic-that would cause
~
increased congestion, impeding both the evacuating-and
. commuting; traffic.
To' illustrate the effects of commuter' traffic, we
.added only 200 additional. commuters to'the'" combined run" reported earlier and the effect was an increase in ETEs_of'10 minutes.
.Since this. represents only a small' percentage of~the 54,000 potential commuter trips, the effectlof explicitly including all commuterftrips would-be substantial.
Q.
You.also listed a potential capacity-related problem with disabled vehicles.
Would you explain this issue further?
A.
Volume 6 discounts the problem of disabled vehicles by asserting (1) that disabling' accidents are unlikely in an evacuation due to low travel speeds and (2) that disabled Lvehicles'could, in any case, "be pushed onto the shoulder" (pp. 12-3, 12-4).
The Lieberman affidavit for summary disposition of SAPL 31 (March 25, 1987) contends further that
' disabled vehicles on bridges or other choke points would either
~be pushed out of the way or would be stored in a location where traffic could move around the disabled vehicle, "even if it means encroaching, somewhat, into the incoming lane of travel on a two-lane road."
In this latter situation, it is asserted that " traffic flow in the inbound direction would take Y-
L advantage of any gaps in the evacuation traffic."
Nowhere does
+
. Volume 6 indicate how many of the' evacuation routes in the EPZ-have adequate shoulders to-store disabled vehicles.
Presumably, this information was gathered in the fiel'd survey referenced in Volume 6 (p. 1-10), but records'from the logs that were compiled by KLD staff in surveying these routes were discarded by KLD (even though'these records are explicitly referenced in Volume 6).
During winter conditions, snow banks frequently are large enough to prevent disabled vehicles from being pushed from the traveled way onto the shoulder.
The assumption that incoming vehicles will find gaps in evacuation traffic that is diverting around such obstructions is unsupported at best given the heavy volumes of both evacuating and of counterblow traffic.
Further comments on this issue appear in Dr.' Ceder's testimony.
In. response to the contention that vehicles removed to the shoulder will reduce effective capacity, the Lieberman af fidad t compares the capacity of 1728 veh/hr/ lane used in the KLD study for the area's freeways to observed flows of 1800 during construction operations when lane widths were reduced (as cited in the Highway Capacity Manual).
The comparison cited, however, is not valid because it does not include the effect of additional capacity reductions from use by non-commuting populations present in large numbers on a weekend l
J evacuation.
The Highway Capacity Manual recommends additional i
reductions between 10 and 25% to account for lower levels of j
familiarity with the road network for weekend and recreational l,
?
{
traffic.
If the 1800 veh/hr/ lane observed capacity figure is adjusted as recommended to account for this effect,.it reduces to between-1350 and'1620, substantially lower than the capacity
.of 1728. assumed in the ETE study.
Neither the KLD study in Volume 6 nor the Lieberman affidavit indicates how long it is estimated for a tow truck to respond to an incident and return to its assigned location and, therefore, how many incidents could be managed by each truck.
The ETE study assumes that low vehicle (average) speeds will result in relatively few accidents which produce disabled vehicles.
However,-Volume 6 assumes an average speed for incoming buses of 40 -50 mph.
Undoubtedly, many commuters will be rushing home to pack up their families at speeds reaching 40-50 mph between intersections.
Thus the potential for high-speed accidents exists.
Even though the inbound bus
" average" speeds will be higher than speeds the buses will commonly use inside the EPZ (because most of the travel outside the EPZ will be on interstate or other high-speed roads) the travel within the EPZ will have to average about 40 mph to maintain the cited overall 50 mph average.
If evacuating vehicles are forced into the opposing lane by vehicles with mechanical or other failures, the potential for an accident with these high speed counterblow vehicles is significantly increased.
There are at least three locations where evacuating vehicles are routed into lanes normally used by vehicles traveling in the opposite direction:
the intersections l
depicted in TCPs B-AM-06 (p. I-19), B-SA-06 (p. I-28) and D-HA-02 (p. I-39).
At all of these locations, a significant potential exists for head-on conflicts between returning
-commuters and evacuating vehicles.
Overall, the likelihood of.
disabling accidents appears to be greater under evacuation-conditions than-in normal circumstances.
See the testimony of Dr. Ceder for additional testimony on this' point.
I have one final point regarding disabled vehicles.
The beach survey' conducted by AEL Associates found that 13% of those surveyed would abandon their car if little forward l
Progress were made in 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> and over 1/3 of the beach-goers would abandon their car if little progress were made in 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />.
Since even the evacuation time estimates in Volume 6 show longer waits than these for certain beach evacuees, prudent evacuation planners must consider the likelihood of l
significant vehicle abandonment.
Q.
Finally, you mentioned the effects of weather on roadway capacity.
Would you explain your findings on this l
issue?
A TOH Contention III Basis B argues that the effects of adverse weather conditions on roadway capacity and, correspondingly, on ETEs are not adequately accounted for in the ETE study.
The ETE study states that " estimates of roadway capacity must be determined with great care" (p. 3-1).
It is further stated that the calculated effects of weather on nighway capacity were " based on limited empirical data" (p.
3-1).
In fact, as argued in this contention, no site specific 34 -
data on the extent of delay caused by these road hazards were collected.
In addition, the effects of fog were simply surmised to be equivalent to the effects of' rain.
The effects of rain were in turn estimated to be equivalent to those observed in studies conducted elsewhere.
The Lieberman affidavit of March 25, 1987 (TOH III),
states that "public officials indicated that ocean fog is an unusual occurrence during the summer months" which " generally dissipates at 9 a.m.
or may appear after sunset."(p. 6) No names of these officials are given.
Other data, however, indicate that ocean fog is common in the summer months.
Further, rain and fog together are not uncommon.
Weather records for Portland, Boston, and for Pease Air Force Base (NH) provided by the Applicant in response to discovery in fact support the contention.
Those records show that " heavy fog" is experienced between 5 and 6 days per month throughout the period of May to October in Portland.
Over that same period general fog was observed at Pease (1957-1979) on 46% to 54% of the days and 18% to 22% of the total hours.
This record indicates a fog presence much greater than " unusual".
In Volume 6 and in the Lieberman affidavit, the discussion of effects of fog asserts that a sudden rain scenario, as modeled with a 20% capacity reduction, is responsive to this issue.
The only empirical evidence cited is from the 1985 Highway capacity Manual, which states that "10 to j
20% reductions [in capacity] are typical and higher percentages are quite possible" (emphasis added).
No data specifically on
(
35 -
1 A
effects of fog are cited, nor was any data on the influence of I
fog on traffic flow along the New Hampshire seacoast collected.
Given the prevalence of this condition, prudent
('
planners would not' fail to obtain empirical observations of the l
effects of fog on the capacity of critical evacuation links in i
the Seabrook EpZ.
Such empirical' observations could be made quite easily.
Further, prudent evacuation planners would not fail to make empirical observations of the combined rain / fog condition, given how easy such observations are to make and how critical this capacity-reduction factor is.
In his affidavit of March 25, 1987 (SApL 31),
Lieberman also states than no estimate of the effect of roadway ice on capacity was found in a litet ?ure review-He then contends that "(i}n the absence of such data, 25% reduction in capacity as used for snow, appears acceptable."
(Lieberman i
affidavit, pp. 19-20).
Lieberman supports his position by stating (p. 20) that, although "[hlighway capacity during an l
ice storm may be less than that during a snowfall", this is offset because trip generation times will be less given that i
1 driveways would not need shoveling.
However, the latter effect will be small compared to the capacity reduction because the j
time at which trips are added to that network is not as important as the number of trips.
Moreover, Lieberman states that "under severe ice conditions, in the absence of sanding, some highway sections with extended upgrades may become 1
l virtually impassable."
(Lieberman affidavit, March 25, 1987 (TOH III p.
7)).
To justify ignoring this condition, l l i
(
.______ 9
Lieberman states:
"It is my understanding that sanding is a prevalent procedure in this area."
Id.
However, in his deposition, Lieberman states (June 29, 1987,
- p. 145):
Q.
Do you have any idea how many sanding vehicles would be needed?
A.
No, because I don't believe that sanding operations should take place during an evacuation.
They're more of an impediment
"' ?
than a help.
Q.
Why is that?
A.
Because they're very slow-moving, slaggish vehicles, and they would slow down traf#'~.
For the same reason avuldn't want
, see snow plows out there during an evacuation.
In any case, sanding and ploding operations would be severely impeded by traffic jams, barriers and cones, causing sand truck cycle times to be substantially longer than under normal conditions and resulting in ice accumulations that would make upgrades " impassable."
This will be inevitable unless additional specific provisions are made in the plan for providing extra sand trucks, beyond the current available fleet, to offset increased cycle times.
Q.
You mentioned earlier that the ETE computations provided by KLD are biased on the low side by as much as 30 minutes because of a conceptual error in the interpolation method they used.
Would you please explain this further?
A.
The I-DYNEV outputs used by KLD to compute ETEs show the cumulative number of vehicles that have passed through a i
given area (e.g.,
the edge of the EPZ) at half-hour intervals.
Since all of the vehicles may finally clear out of an area in i ___--- -- -
between two 30-minute intervals, an adjustment procedure was-used to determine when, during the last 30-minute interval, the area actually cleared.
The method used by KLD to determine the ETEs to the nearest.5 minutes, as reported in Volume 6, is described in the Lieberman deposition of June 29, 1987 (p. 117):
We generally used as a matter of course half-hour outputs.
And then by interpolation, sensible interpolation, we were able to pinpoint what the ETE is, certainly within plus or minus ten minutes and probably within plus or minus five minutes.
This " sensible interpolation" was accomplished by the following method (based on Lieberman deposition pp. 120-121):
30 x (Vehicles 2 divided by Vehicles 1)
Where:
Vehicles 2 is.the change in' total vehicles discharged over the last 30-minute interval in which vehicles remain in the selected area; Vehicles 1 is the change in the number of vehicles that passed through a selected area in the half-hour period proceeding that in which the area emptied out; and 30 is the lenoth of the interval, in minutes.
This method assumes implicitly that all of the roads within this area empty at an even relative rate throughout the final 30-minute interval in the evacuation.
In fact, it may be that traffic remains on only a single road during this last period and that it takes much of that period to empty that one road.
Thus, the correct way to perform this interpolation is to compute the quantity shown above for each roadway in the I
selected area separately.
Since the ETE is defined as the time l
'1 at which an ares is cleared of all vehicles, the ETE adjustment
]
is correctly computed as the maximum of these values.._-_____________________d
A. simple example may be helpful in illustrating this point.
Assume that there are six roads leading out of the area being evacuated, each of which carries 500 vehicles when loaded to capacity during the next-to--.he-last 30-minute interval in the evacuation.
During that period a total of 3,000 cars would leave this area
(= 6 x 500).
Assume, then, that all but one of
~
these roads empties completely during that.next-to-the-last period, and that the final road has 400 vehicles left to e
clear.
Since that road can carry 500 vehicles in 30 minutes, it would take an additional 24 minutes
(=
30 x 400 divided by 500) to clear this final road.
The KLD interpolation method would compute, instead, the ETE final period adjustment to be only 4 minutes
(=
30 x 400 divided by 3,000).
Thus the KLD interpolation would, in this example, underestimate the actual i
evacuation time by 20 minutes as a result of this incorrect methodology.
More generally, the methodology used by KLD introduces a systematic bias that underrepresents actual ETEs by between 0 and 30 minutes.
l Q.
You referred earlier to work that Dr. Luloff conducted q
to estimate and project resident population and employment I
levels.
Would you first explain which contentions / bases you used this material to address?
A.
TOH III/ Basis E and SAPL 31/ Basis 3 both reference the J
concern that the New Hampshire seacoast area's "significant growth" in population and employment will quickly make the ETE estimates obsolete and that the effects of this growth should be anticipated in the current NHRERP.
The December 1986 RAC _ _ _ _ _ - _ _ _ _ _.
p l.
review (" General Comments," #7) recommended that:
If additional significant growth is anticipated, then ideally the impact of future increases in employment on evacuation time estimates should be considered..
(Similarly, if growth in the population of permanent residents is anticipated then that growth should also be considered in determining evacuation times.)
There is a very good reason for requiring that ~ the effects of this growth be considered in this Plan now, even though the growth will presumably be reflected, as it is realized, in future plan updates.
Growth will result in longer-ETEs unless appropriate highway improvements are put into place to offset the larger vehicle loadings.
The planning process required to achieve those improvements is often a lengthy one, ranging over ten years for major projects involving federal funding and environmental impact studies.
For those improvements to be in place in time for.them to avert unacceptable increases in evacuation times, growth must be anticipated several years in advance.
Q.
What did you do to estimate the likely effects of future growth?
A.
Dr. Luloff projected resident population levels for the New Hampshire EPZ towns, as described in his pre-filed testimony and in various tables and written material that he transmitted to Resource Systems Group.
We translated his New Hampshire population estimates into I-DYNEV vehicle loadings using the methodology described in Volume 6.
All other KLD inputs, including the 1985 Massachusetts EPZ town populations reported in Volume 6, remained unchanged.
The resulting ____ ___________ ______________ _______ -
I-DYNEV runs show an 8% increase in ETEs for Scenario 1 (the whole EPZ) for the year 1990 population estimate, an 11%
increase-for 1995, a 24% increase for the year 2000, and a 55%
increase for 2010.
Also, because the Town of Seabrook is growing relatively faster than the other towns in the area, the network bottlenecks are shifted by this growth.
Q.
Is there additional work of this type that should be completed in order to assess the full effects on ETEs of future growth?
A.
Yes.
Growth in employment, which has been even more rapid than population growth (Volume 6, p.
5-1), was not factored directly into our analyses.
Similarly, the effects of' additional retail facilities, hotels / motels and other establishments constructed since the 1981 NRC (Kaltman) study on which the Volume 6 estimates are based, were not considered and properly should be.
The effects of future increases in beach area parking should also be considered.
A comparison of the August 11, 1985, aerial photographs relied upon in Volume 6 with current conditions indicates the formation of new parking lots in the interim 2 years.
This is a trend which will likely continue as population pressures on the beaches escalate and which should and would be anticipated by prudent planners.
Finally, our analyses did not consider the effects of any anticipated future growth in the Massachusetts towns whose traffic mixes with New Hampshire vehicles in many key evacuation locations.
All of these growth effects, taken together, will lengthen ETEs well beyond the levels indicated by our more limited analysis. - - _ _
p f-Q.
Were there;other I-DYNEV runs that you did which L'
showed-significantidifferences'in ETEs from those reported in t-Volume 67 LA.
'Yes. -We conducted numerous other. analyses,' including' one set-which was intended-to test.the sensitivity of the ETEs to input values which.were derived from.the telephone survey commissioned bygKLD.
One'particularlylimportant input KLD obtained from this survey.is.the. assumed number-of residents who will evacuate in each' car.
SAPL Contention 31/ Basis 6' argues-that the KLD survey is potentially_ biased to non-response and that there is "therefore no reasonable basis-for assuming that the... estimates of average person occupancy of vehicles evacuating the-EPZ or other data, derived from the survey are accurate."
Basis 19-specificallyfasserts that "[t]he-estimate of 2.6 people per. vehicle for permanent residents is unrealistic."
This value is importantbecause it directly determines the number of resident vehicles which will be' loaded onto the roadways during the evacuation.
KLD used'a value of.2.6 person / car, which is based in part on the KLD survey results and in part on somewhat arbitrary assumptions regarding likely household behavior (Vol.
6, p. 2-5).
The Social Data Analysts survey included a specific question regarding the number of cars that would likely be used o
in a general evacuation scenario.
Tabulation of those data yields 2.3 person / car.
This value is more reliable than the KLD input because it is derived from a survey with lower non-response bias and it does not depend on arbitrary )
l assumptions by the analyst about likely household behavior.
The testimony of Dr. Stephen Cole describes in greater detail the problems with the survey relied upon by KLD.
When we conducted an I-DYNEV run using 2.3 persons / car, and held all other inputs constant at KLD's values, it resulted in an overall ETE increase of 7% for Scenario 1.
Q:
What other types of ETE estimates did you prepare using I-DYNEV?
A.
We performed _a set of analyses that tested the assumptions about the way the overall traffic management plan would work.
Volume 6 computes ETEs based on a traffic management plan that is described in various sections of that document, including sections 9, 12, and appendices I, J, K and L.
In particular, the evacuation routings are designated in appendices J and K while the other mentioned sections specify how intersections would be controlled to facilitate traffic flow along those routes.
There are many critical assumptions made in these sections of Volume 6, assumptions which are contested in various admitted contentions, including SAPL Contention 31/ Basis 4 ("[it is unrealistic] to assume that traffic management and control measures are in effect at the time the evacuatior. is ordered") and Town of Hampton Contention III/ Bases (C)2 and F.
Q.
What specific tests of this traffic management plan did you conduct?
A.
We evaluated the effect on ETEs of three different types of likely traffic management problems: _ _ _ _ _ _ _ _ _ _ _ _ _ _ -
4 (1)
Delayed' staffing of some-of the traffic control posts (TCPs);
a 4
~(2). Evacuees' choice of host locations and routes;other.than those assumed in the-VolumeL6 analyses;'and
.(3). Non-optimal traffic flow control at'
- intersections.
. ould youfplease_ describe the ways in which'you Q.-
W represented:these effects and'the results that you obtained?
.e A.
SAPL131/ Basis 4 asserts that the ETEs presented in Volume 6 are~not realistic because they fail to-account for
.g delayed staffing of TCPs and'ACPs.
The analyses reported in Volume 6 assume that all transportation.and access control posts are' fully operational at-the time.of the beach closing' message.
See. Volume:6, p. 10-70.
There are at least three locations at-which TCPslare configured to use lanes in ways substantially.different from the ways they are used in normal
~ flow cond'itions.
The I-95/Rt.51/Rt.101C intersection is
. changed by!p;acement of'29 traffic cones to allow westbound.
evacuating Rt.51 traffic to travel in the eastbound lane (TCP D-HA-02, p.
I-39).
(This TCP diagram is Attachment 5 to this testimony.)
This operation depends, in turn, on the prior placement of traffic control measures (see TCP F-EX-04, p.
.I-74) (Attachment 6) on Rt.51 in Exeter to stop any flow of eastbound traffic from continuing to the I-95 interchange.
This eastbound movement is to be physically prevented by placement of a " Truck" at the Exeter TCP.
The net effect of this arrangement is to enable use of both Rt.51 overpass lanes (over I-95) for evacuating traffic, doubling the evacuation..
h_,__,._.,__._,._.___m. _ _ _. _. _ _
u
I-L
. capacity over what would otherwise be available from the single westbound lane..
Similarly, at the intersection of Rts.1, lA and 110 (TCP B-SA-06, p.
I-28) (Attachment 7), an eastbound.
lane of Rt.110 is converted through the intersection into a westbound evacuation lane.
In this case, there are no
-downstream controls specified to eliminate encroachment by eastbound vehicles originating from the numerous curb cuts along Rt.110 west.of this intersection.
Finally, an I-95 southbound off-ramp is' converted to an on-ramp to double 1
evacuating capacity at the I-95/Rt.110 interchange (TCP 1
B-AMN-06 p.
I-19) (Attachment 4).
As noted above, the I-DYNEV results presented in Volume 6 assume that all of the traffic control posts are fully operational from the beginning of the beach closing (which is assumed to occur 25 minutes before the general evacuation).
We i
ran the I-DYNEV Scenario 1 model assuming, instead, that posts at the Rt.110/I-95 and I-95/Rt.51/Rt.101C intersections were not fully operational until I hour and 45 minutes into the general evacuation (or approximately 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> after the beach closing).
The result was a 7% increase in evacuation times over those reported in Vclume 6.
Longer delays in staffing these posts will have proportionally larger effects on the ETEs.
i Q.
Is it reasonable to assume that at the I-95/Rt.51/Rt.101C intersection, the TCP will not be fully I
operational for two hours after a beach-closing announcement?
l A.
Well, as I understand the situation, the Town of Hampton is not cooperating in evacuation planning, and all of 45 -
________O
4 its TCP's, includingLthis'one, are to'be staffed by members of the New Hampshire St' ate Police.
The'most:recent Summary of-Personne1' Resource Assessment for the NHRERP (Aug. 1987)-
indicates on p. 37 that197 state police are~needed for'NHRERP duties from a total of:185 available state-wide.
The State Police do not"have theLcones and barricades'they need'to_ set up this TCP, however.
These are supposed-to be delivered byga truck: driven by personnel from the New Hampshire Department:of Transportation.
Thus,.these personne1'must first report to:
duty, load theLtruck with cones and barricades, then drive;to ea'ch TCP, unload the designated cones and barricades,-then~
drive to the next TCPr and so on.
Also, as I. understand it, each town's TCPs are to'be staffed and equipped in an order of' priority established from a priority list.provided by,the
~
police chiefs of each town.
For Hampton, the highest priority TCPs are the onesoon the beach roads.
Therefore, given-that the state police for this TCP may be coming from some distance, that the early. arrivals will be assigned to the beach areas,-
and.that the TCP can't be established until the cones and barricades arrive, it~seems unreasonable to assume that this TCP would be operational in less than 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />.
Q.
The second traffic management problem that you raised deals with the host destinations and routes that evacuees would use.
Would you describe what you did to evaluate this problem?
A.
The Volume 6 analyses assume that all evacuees will travel to their designated host communities.
This assumption does not appear reasonable for the large number of beach _. _ - _ _ _ - _
transients from the Hampton beach areas, who are assumed to travel to the City of Manchester, New Hampshire, as a host location.
Under summer conditions, a large percentage of these evacuees come f roin areas south of the Hampton beach area.
According to the plan, all Hampton beach transients are to be routed to the west and north out of the beach area toward Mancheecer.
Many are routed west along Rt.51 and Rt.101C.
We did additional I-DYNEV analyses to determine the effect of 'Hampton beach visitors evacuating to locations other than Manchester.
In one set of runs, we assumed alternatively that 1/3 and 2/3 of the Hampton beach area visitors would evacuate directly to the south along Rt.1A, eventually connecting to I-95.
The effects on ETEs were 15% and 47%
increases (over the Volume 6 general evacuation times for a summer weekend midday evacuation), respectively, for the 1/3 and 2/3 routings.
Even if Hampton beach evacuees were successfully routed to the west along Rt.51, it is likely that, at the I-Fe interchange, many would choose to merge onto the I-95 on-ramp rather than continuing west along Rt.51.
In my opinion, it is unrealistic to assume that those evacuees who are from Massachusetts and who would have proceeded )
oriously along Rt.51 will continue along this route to an untamiliar destination to the west when they have an opportunity to access I-95 e.ir.her northbound or southbound.
The travel lanes of I-95 at this interchange will be moving at 65 mph in uncontested flow in both directions.
We tested the effect of 2/3 of the Rt.51 traffic accessing I-95, rather than continuing west l - --- _ --
_---A
s
.q u 9
.along Rt.51.'
The result, as estimatedby I-DYNEV, is'an increase.in'ETEs-for Scenario 1 of 15%, holding all other'KLD inputs' constant.
One' additional ~setEof-I-DYNEV? runs:that we performed
' tested'the' sensitivity.of the ETEs to assumptions-about the routing'of evacuees,fassuming that they in fact choose to travel to their assigned. host communities.. -The analyses presented-in-Volume'6 assume that-evacuees are able to determine the~ fastest route to their intended destinations-(o.
B-1)'.
Unfortunately,ithe shortest route under evacuation conditions may not be the same as the shortest route under-L normally congested. conditions.
It is not likely that individuals fromtoutside the area will know the fastest route to their destination, as Volume 6 assumes, especially when that route is along unfamiliar roads.
Both residents and beach visitors in the Portsmouth and.Hampton,'New Hampshire beach areas are shown in Volume 6 using a route to the Dover / Manchester host sites via Rts.236, 16h and 16 through Maine in a route to_ Manchester via US 4 in New Hampshire.
These routes are circuitous and use roads which are less well known than the direct routes to Manchester and Dover, Route 101 and the Spaulding Turnpike.
We ran I-DYNEV
- for two conditions:
the first assuming that these New
' Hampshire beach evacuees would not use the circuitous evacuation routes through Maine and New Hampshire, and the second assuming, additionally, that Portsmouth/ Rye-area residents would not use these circuitous routes.
The runs _...-f
i showedi7%.and$11% increases, respectively, in ETEs for the
. summer weekend. Scenario 1, holding all otherfinputs constant.
Q.:
'Would you describe.the work-that-you did to evaluate
.the third traffic management problem that you listed earlier, "non-optimal traffic. flow control at intersections"?
A.
-Many of the evacuation routes' merge at various points with'other routes.. Atithe intersections where two or more major routes merge or cross, traffic controls are-specified in I-DYNEV to indicate the allocation of ' time between the competing directions.
These controls are analogous to traffic lights; they provide 1 alternating opportunities for the traffic.
streams at'each of the approaches to move through the intersection.- The I-DYNEV runs used in preparing Volume 6 assume that these intersections are controlled with 75 seconds between successive " green" indications for'a.given approach
(" cycle length").
In addition, each intersection approach is assigned a " green time" ranging between 10 and 65 seconds, selected to optimize the flow through.the intersection.
Since many of the intersections do not currently have traffic lights and since timings for existing traffic lights are different from those used in I-DYNEV, the Volume 6 analysis
.implict,tly assumes that the assigned traffic guides will manage this allocation process.
There are two problems with this.
First, the traffic guides are not instructed as to the proper timing for the intersections that they are assigned to a
control.
Through much of the evacuation, traffic at the major intersections will be queued in lines that extend farther than 1
t the guides can see, and there will likely be a tendency simply to allocate time evenly among all approaches.
Even if the guides knew the optimal timing for each evacuation condition, they will not necessarily have stop watches to gauge the timing.
The second problem with this assumption of optimal signal timing is that many of the intersections, for which the I-DYNEV Volume 6 runs assume this type of signal control, are not designated as TCPs or ACPs and thus are assigned no traffic guides.
There are at least a dozen such cases within the EPZ; for example, the intersections along Mill Road in Hampton and North Hampton.
In the absence of traffic guides, it is extremely unlikely that the optimal signal assumptions will be observed.
In my opinion, these two problems will, beyond any reasonable doubt, cause the ETEs in a real evacuation for some regions within the EPZ to be longer than KLD calculated them to be.
1 l
We ran I-DYNEV assuming signal timings would be approximately equal for all approaches, again holding all other 1
inputs constant, and ETEs increased 45%.
However, because I-DYNEV does not allow signal timings to vary throughout the evacuation, we were unable to test directly the more likely effects of non-optimal signal timings on the evacuation times of specific subregions under the general evacuation scenario.
For example, the intersection of Lafayette Road (Rt.1) and Atlantic Ave. (Rt.101D) affects the evacuation time for the 5 mile ring.
If the westbound Rt.101D traffic is given more than the 20 seconds (out of a 75 second cycle) allocated in the._
Volume 6.I-DYNEV run for Scenario 1, evacuatior, out of the 5-mile ring along Rt.1 northbound will be delayed.
Generally,
-since specific timings are assumed in the Volume 6 runs, prudent planners would put mechanisms'into place ensuring that they are implemented in the plan.
There are no such mechanisms currently.
Without these, there can be no reasonable assurance that ETEs as short as KLD has projected can ever be achieved.
Q.
You referred in your summary to an'I-DYNEV run.that you performed which included together several of the effects which you have discussed separately above.
Nould you describe this run?
A.
I wanted to test the effect of several.of the factors which describe the conditions of a highly plausible evacuation scenario.
These factors included (1) the more realistic beach parking estimates from the work of Drs. High and Befort, (2) a more realistic estimate of I-95 on-ramp capacities, (3) the effects of a realistic proportion of Hampton beach area visitors heading to alternative destinations, (4) the effect at selected locations of returning commuters, and (5) the effect of delayed staffing of TCPs where traffic was routed into opposing lanes.
We completed this run for Scenario 4 (summer weekday with rain) and the result was a 26% increase in ETE for Region 1 (an evacuation of the entire EPZ):
instead of taking 9:45, as KLD has estimated, we found the ETE to be 12:15.
Again, I want to stress that other than changing I-DYNEV to include these ve factors, all other I-DYNEV inputs remained the same. _ _ - _ _ _ _ _ _ _ _ _ _
When we tried to complete an analogous I-DYNEV run for Scenario 1 (summer weekend), however, I-DYNEV produced a run-time error indicating overflow of an integer variable.
Since we had been denied access to the I-DYNEV source code, despite our request for it, we could neither check on the source of this model error nor could we correct it.
We were able to complete the run only after removing the inputs that specified delayed staffing of TCPs.
Even without this effect, however, the ETEs were dramatically different from those reported in Volume 6.
Evacuation times for Region 1 (the entire EPZ) are over 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br />, almost double what is given in Volume 6.
Similarly, ETEs for other regions are substantially higher that the Volume 6 values.
We simply do not know how much higher the ETEs would be if I-DYNEV were capable of-computing this run with the additional input factor we were i
forced to leave out -- the delayed staffing of TCP's.
Q.
Did you conduct I-DYNEV runs other than the ones that you have described so far?
A.
Yes.
In total, we ran approximately 75 different i
simulations.
Figure 4 (next page) summarizes the results of the runs described earlier. to this testimony provides additional details on these runs.
We ran several i
tests to simply validate the I-DYNEV results reported in Volume i
6 and numerous others to test ETE sensitivities to other l
l inputs.
We did not formally audit the full set of runs listed i
l l
in Volume 6 for internal consistency and for consistency with i
the TCP and ACP diagrams (Appendices I and L) and with the
{
1
(
I
! L__________
j_
routing instructions (Appendices J and K).
We did do some spot checks and found an inconsistency with-Rt.101 traffic routing.
Q.
What was that inconsistency?
A.
According to the routing instructions in Appendix J, all those who are routed along Rt.101C are directed onto I-95 north at the I-95/Rt.51/Rt.101C interchange.
The TCP diagram for this intersection (Appendix 5) in fact shows barricades
" discouraging" westbound movement along 101C beyond the I-95 interchange.
However, the I-DYNEV simulation for Scenario 1 (summer weekend) used in Volume 6 actually assumes that some traffic continues west along 101C through this intersection, despite the barricades and despite the routing instructions.
Q.
Did you find any other discrepancies between what was modeled and what was reported?
A.
Yes.
We also found that the input file provided to us by KLD, which they describe as the input to the Region 1/ Scenario 3 I-DYNEV run (an evacuation of the entire EPZ on a summer weekday), gives different results than those reported in Volume 6:
the inputs that we were given result jn a Region 1 evacuation time 40 minutes longer than reported in Volume 6.
The number of inputs required to make each of these runs is substantial, and it is not surprising that there are some discrepancies.
The prudent and appropriate procedure for verifying modeling which is the basis of any important planning project is to commission an independent audit of the work.
In this case, independent experts in evacuation planning, traffic l
l _________-_____-_D
- _ _ _ _ _ - _ - _ =.
I engineering and computer modeling should verify all important inputs and determine consistency between as-modeled (in I-DYNEV), as-described (in Volume 6), and as-exist (in the actual road network) conditions.
To our knowledge, no such comprehensive review of the work reported in Volume 6 has been completed; if it had, at a minimum the discrepancies reported above would have been identified.
The fact that no such audit has been performed on a modeling project as important as this one -- where lives are at stake -- represents a significant deficiency of the current Plan.
Q.
Volume 6 describes, in effect, a transportation management plan for the evacuation, and you earlier commented on some specific deficiencies of that plan.
Do you have a professional opinion as to the " workability" of the overall transportation plan?
A.
Yes.
The plan in my professional opinion is unworkable.
I have strong concerns about the ability of returning commuters, buses and emergency vehicles to pass quickly and safely through the Access Control Posts and Traffic Control Posts, as they are currently structured and staffed.
The specific contentions addressed by these concerns include Town of Hampton Contention III/ Basis (C)1
(" gridlock between evacuating private vehicles, commuters and emergency vehicles attempting to enter the EPZ"), SAPL Contention No. 31/ Basis 7
(" commuters will be returning home partly against the direction of evacuating traffic, through intersections with cones blocking desired turning movements") and SAPL Contention No. 37 !
-_________-______________-__--__-A
.("no assurance that effective use of these vehicles will be possible'in view of a potential outgoing flow of evacuating traffic").
My concerns relate to these contentions and are based on several observations and analyses.
First, I am concerned that substantial delays to vehicles legitimately attempting to enter the EPZ will occur at Access Control Posts.
According to Volume 6 of the NHRERP (p. 9-1),
Entry (through the ACPs] should be permitted for the following groups:
l Commuters returning to the EPZ, to gather members of their household for the purpose of evacuation.
Transit vehicles (buses, vans, ambulances) dispatched to the EPZ to participate in any evacuation.
All vehicles transporting emergency response personnel..
All other travelers seeking to gain entry to the EPZ should be denied access and provided with local diversion routes.
(emphasis added)
In order to ensure that "other travelers" are " denied access," New Hampshire Radiological Emergency Response Training Module 13 (Traffic Management) states that access control managers should allow entry of " emergency response personnel with reasonable identification of the agency they represent members of the press with press credentials
. area residents with a bona fide need
. residents must show appropriate identification.
(emphasis added]."
Clearly, these people must be stopped at the access control points to determine that appropriate _ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _ _
a
screening tests are met.
The latter group alone (area residents)-will include, according to data in Volume 6, up to 23,000 vehicles driven by EPZ residents who work outside the EPZ and will return to secure their homes and collect family.
- members.
This will result in a need for over 600 resident vehicle identification checks at each post, even if they are spread uniformly across the ?8 access control points.
We used data from the Social Data Analysts' survey to determine the likely number of vehicles that would pass through particular critical Access Control Posts.
According to these data, the I-95 Post GT-2 (Vol.
6, p. L-5) may need to process up to 5500 returning commuters' vehicles.
Assuming that a vehicle identification check requires only 15 seconds on average, almost 23 person-hours of control manager time will be required at this point just to accomplish this function.
Since only 2 traffic guides are stationed on I-95 at this location, this would amount to a delay of over 11 hours1.273148e-4 days <br />0.00306 hours <br />1.818783e-5 weeks <br />4.1855e-6 months <br /> in total to these vehicles.
Even if identification checks were entirely abandoned, all of this traffic would still be delayed because it is routed via barricades into a single travel lane.
At ACP NN-2 (Vol.
6, p. L-29) on the Spaulding Turnpike in Newington, New Hampshire, almost 1900 commuter vehicles (not counting those from Pease Air Force Base) could be returning into the EPZ.
With two of the three guides assigned to this location doing nothing but vehicle checks, over 4 and 1/2 hours will be i
required for this function.
l 1 _ _ - _ - _ _ _ _ _ _ _ _ _
~M o'
?
iD V
The effect of vehicle' identification. checks andfof-1aneLrestrictions'for vehicles attempting'to enter the EPZ will
- be to substantially delay the evacuation of. residents by delaying 1the return of commuters.
Equally'important, there will;be long queues of. vehicles ~ awaiting identification checks, or simply. waiting to pass;through'at the ACPs. 'Such' queues will: interfere significantly with-the movement of emergency; E
vehicles ~and buses attempting to reenter the EPZ and could also directly ' impede evacuating traffic' flows.
Another serious problem with the traffic management plan concerns'the way in which cones and barricades will be.
placed=at ACPS and TCPs.
Returning commuters, emergency. buses, Land other emergency vehicles traveling in directions other than G
outbound along the. primary evacuation routes will be significantly impeded by these' obstacles.
At most of these
~
control points, each lane through which movement is
" discouraged" is blocked with at least 3 cones (see,'for-examp1'e, ACPLNW-1 on p. L-28).
(Attachment 9)..The maximum roadway lane widths cited in. Volume:6 (p. 3-7) are 12 feet (some lanes are as narrow as 10 feet).
If two of the cones are~
placed at the outer edges of the lane and the third is placed at the center of the lane, as shown in Volume 6 (see Attachment 9), clear' aisles.less than 6-feet wide remain.
Six feet is barely enough space for a passenger car; school buses and maintenance vehicles are up to 8 feet wide and thus would not fit through these aisles, nor would a passenger car that must turn through the intersection.
Instead, these vehicles must stop while the cones and barricades are moved.. __ _- -____ _ _ - _ _ _
Furthermore, barricades are specified at several l
locations where placement according to the schematics provided in Appendices I and.L would physically obstruct traffic flowing into the EPZ.
For example, TCP D-HA-02-(p. I-39) (Attachment
- 5) at the intersection of I-95/Rt.101C/Rt.51 in Hampton shows i
four barricades across Rt.101C.
Given that Rt.51 is blocked in Exeter to westbound traffic by barricades and a " truck,"
Rt.101C serves as the primary route into the EPZ from the west and is, in fact, a designated bus route.
Volume 6 states that
)
all barricades will meet MUTCD specifications (p. 9-14).
These specifications, listed on p. 9-19 of Volume 6, state that the minimum allowable rail length is 2 feet (4 feet for Type III).
If these four barricades are placed as shown in the TCP D-HA-02 diagram (Attachment 5), there would clearly be insufficient room for either cars or buses to proceed westbound along Rt.101C as would be required for orderly movement of these incoming vehicles.
A more general overall concern that I have with the descriptions of traffic control strategies at the ACPs and TCPs is that there is substantial ambiguity and, in some cases, seeming contradiction as to how the controls are expected to work.
In some cases, such as the one cited above at the I-95/Rt.101C/Rt.51 intersections, traffic is apparently to be
{
l allowed past barricades.
This is consistent with the policy stated in Volume 6 (p. 7-1):
We employ the terms " facilitate" and j
" discourage" rather than " enforce" and
" prohibit" to indicate the need for
]
flexibility in performing the traffic control l
I.)
function.
There are always legitimate reasons for a driver to prefer a direction.other than that indicated.
However, there are several locations where allowing a driver to move in a " direction other-than that indicated" will result in a' head-on conflict with evacuating traffic.
For i
example, a " movement discouraged" indication is given on TCP V-AM-06 for the I-95 southbound off-rump traffic (p. I-19)
(Attachment 3).
However, as,noted above, this off-ramp has been converted for use as a primary evacuation on-ramp.
If the off-ramp traffic is not prohibited, a direct head-on conflict will be created.
Similarly, at TCP F-EX-04 (Rt.101C/Rt.51/Rt.101D in Exeter), movements from Rt.101C and from Rt.101D onto Rt.51 eastbound are " discouraged" )-(p. I-74)
(Attachment 10).
However, unless these movements are prohibited, a head-on conflict with evacuatino traffic will be created on Rt.51 at the I-95 interchange (p. I-39) (Attachment 5).
These head-on conflicts will, at a minimum, be disruptive to traffic flow and will, additionally, create substantial safety risks.
Q.
What will be the effects on ETEs of all of these traffic management problems?
A.
It is difficult to say precisely what the effects will be, but I can say qualitatively that the effects will include (1) substantial delays to emergency vehicles, buses and commuters who are legitimately attempting to enter the EPZ, (2) i substantially increased risks of vehicular accidents, particularly at key bottleneck locations, and (3) reduced 1
j l
l efficiency of evacuation flow.
None of these effects have been modeled directly in the analyses presented in Volume 6.
Q.
Have you had an opportunity to review the analyses i
presented in Volume 6 related to provisions for transit-dependent individuals and the corresponding contentions?
A.
Yes.
SAPL Contention 31/ Basis 6 asserts that the telephone survey data used by KLD to develop estimates of transit-dependent populations is biased by non-response and therefore the data are not reliable.
Basis 12 criticizes other data used by KLD to determine bus mobilization times, Basis 13 specifically rebuts the KLD estimates of transit-dependent population; and Basis 17 states that KLD's adjustments for out-of-service vehicles were applied incorrectly.
SAPL Contention 37 asserts that buses (and other emergency vehicles) will be significantly impeded in their trips into the EPZ.
KLD's analyses begin by using the First Market Research telephone survey data to determine the number of people in each town who either do not own a car or who do not have access to a car at the time the evacuation is ordered.
These numbers are increased by 6% to account for vehicles which l
may be out-of-service and are further increased to account for statistical sampling error.
The resulting estimates of people requiring tranait are then reduced by half to account for the assumption that 50% of the earless evacuees would find someone with whom to carpool.
Bus loading assumptions are used to compute the required number of buses and, finally, ETEs for these buses are computed based on KLD's estimates of bus _ _ _ _ _ _ - _ _ _ _ _ - _ _ - - _ _ _. _ _ - - - _ - _ - _
i L
. mobilization times!and route trave 1' times.
Q..
.Do:you'believe that the Volume 6' estimates of transit dependent populations are reasonablyfreliable?~
A.
. tha. - As stated clearly in Dr.1 Cole's. testimony, the First Market Research survey,'from which KLD drew primary data, is generally unreliable.
Further, the KLD analysisimakesi several unsupported: assumptions about the conditions under.
~
which cars would tua availableifor use inLan evacuation'.
In-
' addition to these general prob 1' ems, the adjustment.for out-of-service vehicles was applied incorrectly, as-asserted in SAPL.31/ Basis 17, and the assumed extent.of carpooling is not substantiated.
Briefly, the' adjustment factor for out-of-service vehicles was applied in Volume 6 by increasing the transit-dependent population by 6%.
The stated assumption (p.
11-8) is that "the average. Vehicle is out of service three weeks out of the year."
With this assumption, KLD derives its 6% adjustment factor as follows:
3 divided by 52 = 0.058.
In fact, if 6% of the vehicles were out-of-service, as assumed, the'effect on the size of the transit-dependent population would be much larger.
A simple example illustrates this point.
Assume that a town has a population of 15,000, that an average household size is 3 people (5,000 households), that 1,000 households have no cars, and that the rest (4,000 households) have 1 car (and that that car will be at home at the time of the evacuation).
With no vehicles out of service, 3,000 people will be transit dependent
(= 1,000 hhs x 3 l
l H
li
~
p
(
l M;
' q..
h
'per/hh).
If_5%'of.the cars owned in the town were out-of-service, an additional 200 households-(=.05'x:4,000 hhs) orc 600 individuals will be transit dependent, a-20%
}
increaseLover the-normal transit-dependent population.
The; method used in Volume 6 would increase'the transit-dependent popu14 tion by only 5%, incorrectly assuming that the l'
- out-of-service f raction can be applied directly to the l
transit-dependent population.
In his March 25 affidavit (SAPL 31), Mr. Lieberman defends the " deliberately selected. conservative figure of 6 percent,with which to estimate persons-needing transit because of out of service vehicles" as exceeding what "available sources indicate, which is 1.1 percent of the time."
(p.
25)-
He'does not, however, defend the subsequent misapplication of this percentage.
Lieberman offers, as candidate populations for the adjustment, either the "EPZ population who own one vehicle" or the " estimated transit-dependent population" in his evident uncertainty as to which is the more appropriate.
Id.
In. fact, neither is correct.
proper, conservative adjustment for the increase in the transit-dependent population arising from out-of-service vehicles mandates the consideration of the following populations:
(1) those in households with one car available, where all household members generally have access to the car, and (2) those who normally drive in the cars in one-car households where not all members generally have access.
The assumption that 50% of those not having access to a car will carpool is critical in the Volume 6 estimate of __-
h k'
transit-dependent populatio~ns.
Volume 6 notes (at.p'.11-8)
.that nearly 80% of the transit-dependent population' ride-share'd-in the Mississauga,; Ontario, evacuation.
Powever,.the comparability of that observation in terms of'the general
. conditions 1of the evacuation, geographic. distribution of the L
- population,' vehicle occupancy and announced availability or unavailability <of bus transportation'is not forna11y evaluated.
The Seabrook-EPZ is characterized by~a large number of transients in the evacuation stream; the effects.of these
- transients on the propensity to ride-share should be evaluated more carefully.
In general, since the number of buses required is.directly dependent on this calculation, a more reliable basis should be provided.
Volume 6 "vetifies" the transit-dependent estimates'by comparing them to the results of a June 1986-NHCDA survey. That survey however is, at best, at equally unreliable' instrument; the non-response bias from the mailout/mailback survey format renders it unsuitable for.this. enumeration task.
The' Social Data Analysts survey described by Dr. Cole illustrates a more direct method of determining the number of people who would require public transportation in an evacuation.
A simple question was included in the questionnaire:
"How many individuals in your family will need j,
public transportation in order to evacuate?"
We tabulated the i
responses to this survey question and expanded these to the o
full EPZ.-
The results are, generally, a transit-dependent population twice as large as that estimated in Volume 6 (see 63 -
- 'i e
Figure 5, next page).- These data indicate,.therefore,'a need'
~for'at least twice as many buses as contemplated in Volume 6.
-The'effectLof having only-half the number of buses needed would be to substantially increase the ETEs for transit-dependent 7 f
individuals'above1those presented in Volume 6, Table ~10-10, pp. 10-71 through.10-74.
Q.-
Are there other factors that would.also increase ETEs for transit-dependent individuals?
A.
Yes.
In SApL 31/ Basis 12, KLD's estimates' of bus mobilization times are challenged as being overly optimistic.
According to pp. 11-18 of Volume 6,'these estimates of bus mobilization timesLwere' derived from a " telephone survey ~'of-organizations which own and operate buses."
Volume 6' states (pp. 11-19) that based on this telephone survey, 50% of the available buses could be mobilized in one hour.
However, a
- 1987 telephone survey conducted by NHCDA reveals that only?30%
of the-required vehicles could report to staging areas within
- one hour (March 25, 1987 Affidavit of Edward B. Lieberman (SAPL 31), p. 21).
In fact, the " bus mobilization time" study relied upon in Volume 6 was apparently carried out in a haphazard manner.
Concerning the bus mobilization time survey, the
. Response to Massachusetts Attorney General Interrogatory 52 states:
Because of the limited amount of information which was sought, there was no need to develop a formal survey instrument or methodology.
The importance of having accurate bus mobilization times to an accurate estimation of ETEs for transit-dependent
-____________-__A
l 1
j L
people would suggest the need for more formality and reliability in the data collection process.
The Response goes on to state that:
To the best of KLD's recollection, this survey was conducted in the spring of 1986.
Despite difficulty in recalling when this vital information was obtained, the Response-describes how KLD conducted this survey:
While there was no way to check the reliability of these returns, KLD did engage in extended discussions over this matter with each respondent..During the course of these discussions, KLD emphasized the need for accurate and reliable information and always fed back the responses obtained to confirm that the respondent felt comfortable with the information that he or she gave.
KLD also ascertained that the person was either a manager or a responsible employee of the firm to ensure that the information was reliable.
This rather formal sounding engagement in " extended discussions" on reliability should have been coupled with a formal survey.
An equally important problem, discussed in more detail above, is the likely substantial delays and queueing at several of the important ACPs due to vehicle identification checks and lane restrictions.
The further difficulty that buses will have in maneuvering through cones and barricades at each of the many TCPs encountered in each of the routes will further impede inbound bus flows.
However, Volume 6 assumes an average speed for incoming buses of 40 to 50 mph.
No specific basis for this calculation is given, though Volume 6 states that the average inbound travel distance is 30 miles (p. 11-20).
Assuming that _________ - _ -
=-
C W
, ~.
. only'one-fourth ~of that.distanceE(7.5 mi.)L s within the*EPZ~
i and the' remainder (22.5-mi.) is traveled at:55 mph,_the posted
- speed limit: on most major access roads, in order to average mph,1the EPZ' travel would have to be at speeds averaging-40.
mph.
The'-likelihood'of achieving that average speed'inside:the' EPZ,-and~thus of providing' transit-based' evacuation in the times assumed in Volume 6,.is remote given the many impeding factors that will be present under evacuation conditions.
Q.
You have reported in this testimony many examples of changes in.._I-DNYEV' inputs which dramatically affect the ETEs;
.yet"the " sensitivity tests" described in Volume 6'and those reported in Mr. Lieberman's March 25 affidavits show generally only small effects in ETEs.
How do you explain these differences?
A.
Very simply, we evaluated different sets of conditions.
Mr. Lieberman modeled those which he apparently thought were most likely and we modeled conditions that we have
~
demonstrated through this and other supporting testimony are more likely.
Computer models of behavior as complex as an evacuation from the Seabrook EPZ understandably involve many assumptions about values of input parameters.
However, where
- these values are uncertain, it is important to determine whether changes in the assumed values will affect the major conclusions that are drawn from the work.
In the case of the ETE study. reported in Volume 6, there are many assumed model inputs with which there is reasonable uncertainty regarding the "true" value. _ _ - _ _ _ _ _ _ - _ _ _
Volume 6 reports sensitivity tests for only four sets of' conditions (pp. 10-15 to 10-19).
The first set deals with different assumptions about beach population; however, the two
[
additional scenarios described look only at the effect of reductions in beach population to 80% and 60% of the original estimates.
The work reported by Drs. High and Befort indicated that much higher beach populations are both possible and likely to occur.
Two additional sets of KLD's sensitivity tests deal with different rates of accident escalation.
The final reported set of sensitivity runs describes the effects of two sets of highway improvements.
Noticeably lacking are tests which describe the effects of reduced capacity.
In general, the KLD sensitivity tests ignore important issues such as the effect of returning commuters, the effect of evacuees choosing to travel to host locations other than those assigned, the effect of a possible underestimation of beach population, and the effect of spontaneous evacuation outside the EPZ.
The March 25, 1987, Lieberman affidavit cites tests of the effects of "10 road blockages at different locations throughout the EPZ."
These " blockages," however, were specified to be of limited duration and served to reduce capacity over that period but not to " block" flow.
There are also several effects that KLD chose not to model explicitly, arguing instead that they would have little impact on ETEs or that they are outside the scope of the work.
Some of the effects KLD ignored are:
(1) the substantial flow of returning commuters, (2) the delay caused by identification l i
checks, (3) the impedence produced by cones and barricades, (4) the choice by beach evacuees of alternative destinations, (5) the "non-optimal" traffic signal timings, and (6) the area's projected. growth.
Each of these effects has an influence on l
'ETEs and should be evaluated through sensitivity tests such as the ones reported here.
Q.
Have you reviewed Appendix 4 to NUREG-0654/ FEMA-REP-1, Rev. 1 ("NUREG-0654"), which pertains to the requirements of an evacuation time assessment study?
A.
Yes, I have.
I understand that it presents "an example of what shall be included in an evacuation times 4
assessment study and how it might be presented."
(NUREG-0654, App. 4, 4-1).
Q.
Does Volume 6 of the NHRERP, Rev. 2, meet all of the requirements ser forth in Appendix 4?
A.
No, it does not.
For example, in paragraph I.A. (at p.
4-1)., Appendix 4 call for a " vicinity map" which identifies, among other things, " topographical features."
However, tb7re are no maps in Volume 6 which identify topographical features.
In paragraph I.B.
(at p. 4-2), Appendix 4 states:
"All assumptions used in this analysis shall be provided."
As my previous testimony has recounted, during our assessment of KLD's work we uncovered assumption after assumption which was not provided in Volume 6.
One example is the assumption that KLD makes, but does not mention, about returning vehicles -- that they will not impede the efficient flow of evacuating vehicles out of the EPZ at critical intersections.
Another example is that KLD does not explain its assumptions or method of interpolating its data to determine the "end time,"
i.e.,
the ETE, for each of its model runs.
There are numerous important assumptions made by KLD which are not reported in Volume 6.
Paragraph I.C.
(" Methodology") in Appendix 4 (at 4-2) states as follows:
A description of the method of analyzing the evacuation times shall be provided.
If computer models are used, a general description of the_ algorithm shall be provided along with a source for obtaining further information or documentation.
As I have noted, the description contained in Volume 6 of the method used to calculate evacuation times was incomplete.
While a very general description of I-DYNEV was contained in Volume 6, it did not offer enough information for us (or anyone else) to conduct a meaningful assessment.
- Thus, we turned to the " source,"
i.e.,
KLD Associates, to obtain further information and documentation.
While we did obtain a copy of the I-DYNEV system itself, when we requested a copy of the " source code,"
i.e.,
the primary computer program upon which I-DYNEV is built, KLD refused to provide us with this documentation.
In my professional experience, whenever modeling projects have been undertaken which are meant to come under government scrutiny, this primary computer program is generally made available to those reviewing the model and its I
application on a given project.
I assumed that in requiring a
" source for obtaining further information or documentation,"
NUREG-0654 was implicity requiring that this documentation then _ ____ ____ - -
be made available by the " source."
Here, perhaps the single most critica1' piece of' documentation, I-DYNEV's source code, was requested, and the request was refused.
This is not the only documentation we requested which was not available.
As I have noted, KLD discarded all the original documentation from the extensive' highway system field survey it says it conducted.
See Vol. 6, p.
1-10.
It also discarded the sketches it made of each of the August 11, 1985, aerial photographs it examined in counting the beach area parking capacity.
It is not apparent from the slides themselves what KLD counted as a parking space; only the sketches KLD made of these slides would document where it claims these spaces do and do not exist.
Yet these sketches were unavailable to us as well.
NUREG-0654's requirement is meaningless if all it means is that the " source for obtaining further information" must be identified but that no documentation need be provided by the
" source."
Further paragraphs of Appendix 4 to NUREG-0654 pertain 1
to " Demand Estimation" (par. II), " Traffic Capacity" (para.
III), and " Analysis of Evacuation Times" (para. IV).
As I have stated in my testimony, I believe that KLD has significantly underestimated demand, overestimated capacity, and, as a l
result, underestimated ETEs.
For this reason, even if the Volume 6 ETE study met the checklist of requirements set forth in Appendix 4 to NUREG-0654, which I do not believe it does, it z
would not ensure that the ETEc are reliable or accurate to a reasonable degree.
But if that checklist of requirements had 1 t
i been met, then our. review of:the NHRERP's ETE studyLwould'have' been'much. easier.to conduct.
'Let me give1you an example. ;In section'III (B.).of' Appendix 41to NUREG-0654 (at p. 4-5), whichLpertains t'o1" Road Segment Cha'racteristics," there is a requirement.that."[a),
r table'such-as' example-Table 1 shallibe provided indicating all the evacuation route segments and'their characteristics, Lincluding: capacity."
The referenced Table.1'(on p. 4-15 of NUREG-0654) (Attachment 11, hereto) contains a column for
' indicating 1" comments."
A footnote.gives this instruction regarding-what " comments" to included:
" Indicate any special I
conditions that may affect roadway capacity."
In doing our review of KLD's work, We would.have found:
-it? extremely helpful to have had.this information for each' roadway link. 'But nowhere in Volume 6 or anywhere else is this n
.information'provided.
This information was among the data collected by KLD staff when, in August 1985, it drove the entire highway system within the EPZ and recorded the characteristics of each section of highway, including special conditions that may affect capacity.
See Volume 6, p. 1-10.
Because this information was not in Volume 6, we asked KLD for this information and were informed that the audio tapes on which this information was recorded had all been re-used and that no transcript or other documentation of this roadway data existed.
KLD said that they had adjusted I-DYNEV's link capacities accordingly, based on this data, but having so programmed the model, it had no further need for the raw data.. _ _ - _ - _ _ - _ _ _ _ _ _ _ _ _ _ _ _ -
i Thus, instead of providing the detailed road segment characteristics NUREG-0654 requires, KLD provided only a table, in Appendix N.to Volume 6, which lists KLD's estimates of capacity for each link.
For us, this meant that when we wanted to check KLD's capacity estimates for any given link-we had to go out to the EpZ and drive the link ourselves, gathering the very same road characteristic data that NUREG-0654 requires be included in the Volume 6 ETE study.
This made it extremely time consuming to reveiw all of KLD's capacity estimates.
As a result, we drove only those links we found to be most critical, checking for "special conditions" which might reduce capacities.
As I have noted, we found some important "special conditions" which apparently were not considered in KLD'S analysis.
There is another sensible reason for NUREG-0654 to have this requirement that special roadway conditions affecting capacity be listed for each link.
As conditions change on any
- link, e.g.,
as road improvements are made, or as impediments such as construction projects occur, there is no simple way of knowing whether these will affect the ETEs in the absence of "special condition" data.
For example, if a narrow bridge in the EPZ is widened, those responsible for keeping the ETEs up-to-date will not know whether that may affect (lower) the t
ETEs or not (for that town or region), since they do not know whether KLD considered the narrow bridge to be a factor limiting capacity at all when it did the ETE study.
If those I
responsible for the plan had a simple list of each link's l.
b O,
"special. conditions"'affecting capacity as they-_ monitor and E
. record roadway? characteristic changes in the EPZ, they could reference that' list and know whether.t'o update theLETE'for'that town-(or'ERPA').. As it stands now,'they will haveEno idea,Las road characteristics change, whether to update-that ETE.
l
. Prudent' planners.would always have available'the "special condition" data that is absent from'this Plan.
The simple-solution to-this serious defect in the.ETE study is..to require that.the "special' condition" roadway characteristic data be reported in the way required by NUREG-0654,.or any-similar.way, even if this.means re-driving each link in the EPZ evacuation i
network to regather the' data.
Q.
'Do you. feel that you have been able to complete a thorough evaluation of the reliability of the data and analyses on which Volume 6 is based?
A.
No.-
We have been able to evaluate many, but not all, of the data and: analyses reported in Volume 6.
An initial-necessary test of the validity of any computer-model-based study is that the results can be independently derived and verified (see, for example,. Meadows and Robinson, The
' Electronic Oracle:
Computer Models and Social Decisions, John Wiley & Sons, 1985).
Documentation should be sufficient to allow an independent analyst to audit the trail leading from the data and assumptions to the actual inputs used in the study.
Moreover, the computer model itself should be available for full scrutiny and should be documented to the point that the basis for each key model equation is described and its.;
o computer language implementation (source. code) is revealed.
<The.model outputs should also.be available for comparison to independently _ derived outputs.
The information.provided in Volume 6' falls far'short of this necessary level of.
' documentation.
This is understandable since the purpose of Volume 6 is' simply to report results of the study and to point the reader to supplemental sources of data.
However, even
.after obtaining responses.to interrogatories directed
.specifically atjobtaining the information necessary for our work, we found significant gaps in the audit trail, for example:
Records of KLD field data collection efforts documenting the critical physical attributes of the area's roadways, including both audio cassettes'and typed transcripts referenced-in Volume 6, have been destroyed by KLD staff; No written records were maintained of a survey' conducted by KLD to determine'the availability of buses for evacuation of transit-dependent individuals; We.were denied access to the computer source code on which I-DYNEV'is based.
This meant that we were unable to. verify that the model correctly implements the methodology described in reference documents.
It also meant that we were unable to determine the source of " suspicious" model behavior such as a seeming over-representation of ramp capacity under congested conditions and a run-time error which prevented us from running one of our key model simulations; We were given conflicting data by KLD; the computer input file that supposedly was used for the Scenario 3 run reported in Volume 6
-does not produce the output described therein.
These gaps, among several others, impeded our efforts, and would impede anyone's efforts, to fully critique the work. _ _ _ _ _
I
)
.i I
I f
i More importantly, they substantially diminish the ability of l
anyone other than the same KLD staff who produced the current 1
version of Volume 6 to update the work-in later years, as required by NUREG 0654 For example, as noted above, if future road improvements are made to remove existing bottlenecks, it will not be clear whether those bottlenecks were considered in KLD's rating of a particular roadway.
Or, if a bus company goes out of business, it will not be known whether that company was relied upon in KLD's survey.
Further, others without access to the I-DYNEV model's source code will likely be unable to extend the runs reported in Volume 6 and will not be able to determine the source of model anomalies.
Q.
Do you believe that Revision 2 of Volume 6 accomplishes its stated objectives?
A.
No.
As stated on p. 1-1 of Volume 6:
This plan is designed to protect the health-and safety of the public in the event that an emergency evacuation'is ordered as a protective action in response to an accident ut Seabrook Station.
The material contained in Volume 6 undoubtedly represents a significant amount of effort expended by KLD and its subcontractors to address this objective.
However, the insistence throughout that document that the ETEs produced reflect " conservative" assumptions is challenged by the data and analyses presented in this testimony.
Generally, the assertion that ETEs can be estimated with a reasonable degree of certainty from the data and analyses presented in Volume 6 is not supported.
Further, the transportation management plan _ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _
has significant deficiencies which will reduce the efficiency of'the evacuation and which could, in practice, expose evacuees to risks of serious traffic accidents.
There are.three major sources of error or uncertainty in. estimation of the evacuation times reported in Volume 6:
in calculations make by I-DYNEV, in the quantitative values of inputs to I-DYNEV, and in the overall assumptions made about the behavior of evacuees and of evacuation personnel.
This testimony has dealt primarily with the second of these three sources, the values of inputs to I-DYNEV.
In some cases, such as for returning commuters, we have described effects which could be more precisely quantified given additional data collection and/or analysis.
In the case of the beach population, we have defined, by re-estimating the number of parking spaces, a wider range of vehicle populations that could be present on any given summer day.
Since no one would know precisely how many beach visitors actually occupy the beach area on the day of an emergency, there would correspondingly be a wide range of uncertainty in evacuation times for this
-l population.
This uncertainty could be eliminated only by
{
conducting daily aerial overflights of the area, but it could l
be reduced substantially by compilation of a statistical data i
base that would correlate actual beach counts with weather l
i conditions, daily / weekly trends and yearly growth factors.
{
i of the other two major sources of uncertainty in the evacuation time estimates, it is my professional opinion that behavioral assumptions are more significant than is uncertainty t_1__________
J
result'ing from the methods used.in I-DYNEV.
Having conducted extensive research on travel behavior and having reviewed the testimony of Drs. Zeigler, Johnson, Cole, and Luloff I am concerned about extrapolating, as the plan implicitly does, very-simple models of human behavior to situations that are well outside the realm of common experience.
My fundamental concern is that for those in the-beach areas of the EPZ the evacuation times presented both in Volume 6 and in this testimony are so long that the Plan's simple assumptions.about an orderly evacuation may no longer apply.
Certain1y, the evidence provided by KLD does not substantiate these dubious assumptions and thus does not provide reasonable assurance that an orderly evacuation, under these circumstances, would occur.
Q.
Is there any other testimony that you would like to present at this time?
A.
No.
Thank you.
1 l
i j
l L
I I
b_____