ML20082D035
| ML20082D035 | |
| Person / Time | |
|---|---|
| Site: | Shoreham File:Long Island Lighting Company icon.png |
| Issue date: | 11/18/1983 |
| From: | Pigozzi B SUFFOLK COUNTY, NY |
| To: | |
| Shared Package | |
| ML20082C880 | List: |
| References | |
| ISSUANCES-OL-3, NUDOCS 8311220298 | |
| Download: ML20082D035 (56) | |
Text
- 4 >
l l
UNITED STATES OF AMORICA l NUCLEAR REGULATORY CCMMISSION l
l Before the Atomic Safety and Licensing Boarf
)
In the Matter of )
)
i LONG ISLAND LIGHTING COMPANY ) Docket No. 50-322-OL-3
) (Emergency Planning)
(Shoreham Nuclear Power Station, )
Unit 1) )
)
Direct Testimony Of Bruce William Pigozzi On Behalf Of Suffolk County Regarding Emergency Planning Contentions 65 and 23.D (Evacuation Time Estimates) 4 Q. Please state your name.
A. My name is Bruce William Pigozzi.
p Q. What is the purpose of this testimony?
A. The purpose is to address Contentions 65 and 23.D which concern the evacuation timerestimates present d in the LILCO Plan (see Table XIV,"' Appendix A at V-3).
In particular, I will
-.( -? ,
discuss how those e,atimates, and the model on which they are
. . ' ',' / '
based, are flawedsand founded on unrealistic:ashbmptions.
Fur-f '
, ; L thermore, the time estimates do not account for several factors
-/ k~ /'
which, if considered, would lengthen the time required to t
s r
'e n . f y-7f
- " ". 8311220298 831118 53 ~
PDR ADOCK 0500032a' ena ,:i
+ ._ . .
.= ..
evacuate the EPZ. Thus, it is my opinion that LILCO's evacuation time estimates are inaccurate and unrealiable. In fact, LILCO's estimates should be far higher than reflected in Table XIV.
I. Backaround Q. Please state your occupation.
A. I am an Assistant Professor of Geography at Michigan State University, East Lansing, Michigan. I teach quantitative methods and mathematical modeling to undergraduate and graduate students from my own department as well as from Civil Engineer-ing, Urban Planning and various other units. A copy of my vitae is attached to this testimony (Attachment 1 hereto).
C. What experience do you have in the area of mathematical modeling?
A. Besides teaching mathematical modeling, I have spent a large portion of my professional career working with and developing spatial interaction models. A spatial interaction model measures, simulates and/or predicts interaction between geographic locations. Interaction models may take several forms, including regional econometric models, input-output models which deal with economic interaction, and transportation
.m ..
models to characterize a few. The KLD model used to determine the evacuation time estimates in Appendix A is composed of submodels which are, in essence, interaction models.
As my vitae indicates, I have had experience with trans- l portion modeling. For example, as a regional planner, one of my chief concerns when studying a region is to evaluate the j transportation system available to it. In 1981-1982, I worked with the Michigan Department of Transportation on the " Urban Systems Model." This computer model was developed by Alan M.
Voorhees and Associates and is designed to predict, among other things, traffic flows and patterns. I have also been in charge l of modeling other transportation systems, particularly as Director of Rail Planning for the State of Indiana. In that 4 capacity I developed a computerized model to simulate the operation of the State railroad system.in order to assist the State of Indiana to evaluate the operational and planning strategies open to it. A model for rail systems such as I developed for Indiana does not differ significantly in analyti-cal purpose from the traffic model used by KLD, both are designed to ant'icipate the results of policy decisions.
Q. How did LILCO derive the evacuation time estimates set forth in Appendix A?
1 se 7 , mN4- 3 w -m s wg g- .--p g e y e
.= .,
A. The evacuation time estimates were aerived by LILCO's con-sultants, KLD Associates, Inc., through the use of a computer model.
Q. Please describe this model briefly.
i A. KLD's model is composed of a series of submodels. These
! submodels are linked together and supplemented with judgmental inputs and manipulations. In general, the planning process used by KLD to model and control evacuation traffic from the EPZ and thereby derive time estimates, is a conventional trans-l portation planning process used for urban and regional trans-portation strategies.
i Section III of Appendix A sets out, in a very basic form, the structure of the model and its basic inputs which are di-vided into the following components:
Trip Generation I
-
- Trip Distribution l
- Trip Assignment l Traffic Control (Simulation) f 1
.a . > ,
The Trip Generation process in this evacuation context is simply a function of determining the number of people who will be evacuating, thereby determining how many trips will be gen-erated which the network must accommodate. As can be seen from Exhibit 1 of Appendix A (Figure 1 hereto), the trips originate f rom the origin nodes scattered throughout the network. The origin nodes represent areas of populations. It is common for regional transportation models to generate trips for a number of purposes: home-to-work, work-to-home, home-to-shopping, work-to-shopping, and shopping-to-home. KLD's model, however, considers only evacuation trips generated by the resident popu-lation (and the employees of major employers such as Brookhaven National Laboratory and Grumman) exiting the EpZ. No work-to-home, school-to-home, or other such trips are considered in the KLD trip generation process.
The Trip Distribution function determines the destination for each origin node, although it does not determine what route the traffic will take. The destination assigned for each source node is not the ultimate destination of the evacuees (such as the homes of family or friends), but rather the point at which the evacuees from a particular source will exit the EPZ. The result of this process is a " trip table" which defines the origin and destination of traffic. Exhibit 2 of
1
}
f\ n, s '
~f
, y :
[ $ 5 flI. . hh Iii '3 I
/
,g. ;
h
. c\'
(' N
=
oo***
s
)
e
,e g,
\
a g
[o t
~j q' - -
w ,. .
i s.
_- . *.W,
,, 1 u~~s j
- q.
, yd .:11 <
- a g( ,
~J .,,. $ .Q.
., . Q,. .
. a . i ,, .
,, g-
= 3, ib .// ,
'.. .e s, 4 p*$
~-
- y. .
I ,* 1 g
'y f
f ',!!. '+. h
/ ,
I i
p, I
l
- g. ~
b Il 4 5
g ,
l Appendix A (Figure 2 hereto) displays the results of the trip distribution process for the Shoreham EPZ. " Desire lines" connect each source node to its respective destination (exit node).
In conventional applications, trip distribution is deter-mined by an examination of the various measures of "attrac-l tiveness" of a destination. A variety of mathematical designs, typically of the gravity or entropy maximizing type (defined in discussion of Contention 65.C.4, infra) are usually used to de-termine this distribution. However, KLD has chosen, for its evacuation model, an arbitrary method to determine destinations which will be discussed in further detail below.
l The Trio Assignment Submodel evaluates and " balances" the aggregate' transportation supply and demand situations and assigns gross flows to paths on a capacitated network. In es-sence, this portion of the model considers simultaneously the population to be moved and the transportation network over which the population will travel. The output of this sec ion includes the assignment of trips from each source node to particular routes in a manner which will optimize travel times to the exit nodes specified in the previous step. That is, the model assigns traffic along routes from source nodes within the various 19 zones to exit nodes outside the 10-mile EPZ.
7 y - -
--m a w - w -
5 &
~ =
of
(
si l 5 f 1 E s j!
!jf ,
I I$! !
'o*
,,s, . I j :g 3 .
/
sem,,..
e 24 ?
{\
- ' P
,..".,,~['!,Y' 7 ..
..z2. h ,. :. .
Q. 4 3n .. I 1 7'%
,x
's x...
- s. . ...=.
t s
+.
*' e -s,
; 't
~ ., .
.,__ ; J ,* s x , . -
.I se w *e ,
+,3 7
. ,'1
. N'f
,k e se gd
,14 -t!;y J 'e_= " ~ ~-..a : h # ,i
. . 2,
.p': ,
'4 , .
I I I
~,
y4 i 1
. =
l l
The Traffic Simulation Submodel (DYNEV) Gimulates the evacuation traffic. It considers the entry rate of traffic from the EPZ residential areas onto evacuation routes, the capacities of the roadway sections and intersections of the designated evacuation network (including the effect of special traffic control measures), and the destinations and routes . specified in the Traffic Assignment model. In general, the functioning of the model can be viewed as the balancing of supply and demand. The trip generation and trip distribution functions specify the evacuation demand (i.e., how many trips will be generated to which points). The trip assignment function attempts to satisfy that demand by placing the traffic along the network structure in such a way that all available capacity (supply) is put to optimal use. Finally, the traffic control submodel permits adjustments to the network structure (in response to demand) to reduce evacua-tion times as much as possible. O. Do you have an opinion with respect to Contentions 65 and 23.D? r A. Yes. As Professor Herr has testified (See Herr Testimony 1 on Contentions 65 and 23.D), evacuation time estimates must be as accurate as possible in order to enable those in command and
control of a radiological emergency response to choose between evacuation or some other form of protective action. Evacuation time estimates thus are important to an assessment of whether I evacuation constitutes a viable protective action under various accident scenarios. If evacuation cannot be achieved prior to arrival of the radioactive plume, persons on evacuation routes may be exposed to radiation at the very time they are at-tempting to escape from danger. Thus, time estimates are also a means by which the feasibility of implementing particular protective actions can be assessed. In order to be accurate and thus reliable, evacuation time estimates must be derived from a modeling process based on re-alistic assumptions and complete input data. LILCO's evacua-tion time estimates, however, are not based on complete input data. In particular, the DYNEV model, as used by KLD, has failed to account for the real world circumstances and behavior likely to prevail during a radiological emergency. If KLD had I taken a more realistic approach to modeling an evacuation, its l l evacuation times would, in my opinion, be substantially higher than those described in Appendix A. l W l t l l l
II. Contentions 65.A and 65.B i Q. What is the concern expressed in Contentions 65.A and 65.B? A. The concern expressed in Contentions 65.A and 65.B is that KLD has failed to make accurate and reliable estimates of mobi-lization times for the population -- that is, the period between the first warning message to the population and the time that the population commences the evacuation trip.1/ The particular concern I will address is KLD's failure to account for the large number of pre-evacuation trips within the EPZ which'are likely to occur between the time an evacuation is recommended and the time that persons leave their homes in an attempt to evacuate. Rather, KLD has arbitrarily assumed that there will be a 20-minute " mobilization time" following a rec-ommendation to evacuate, after which all evacuation traffic would begin to evacuate over a period of about 2 hours. I believe the KLD assumption is in error. 1 1/ The County's use of the term " mobilization time" is thus l broader than KLD's definition. KLD states that mobiliza-l tion time is a 20-minute period between notification and l the departure of the'first evacuation trip. (Appendix A, Rev. 2 at V-7). s D s O t I Q. What are pre-evacuation trips? l A. As Professor Herr has described in his testimony on Con-t tention 65, these are the trips that people may take to prepare for evacuation, or for other purposes, prior to evacuation. They include travel from work-to-home, from home-to-school (and .i return), from a shopping area to home, and from home to stores or banks in order to get supplies and money to sustain'the fam-ily unit while relocated. Q. What is the effect of having failed to account for pre-evacuation trips? A. There may be congestion on evacuation routes due to pre-evacuation trips which will delay persons actually attempting to evacuate, thus extending evacuation times. Further, these pre-evacuation trips will also, apart from congestion, cause 1 delays in the time people are ready to commence actual evacua-tion. Also, these trips represent additional demand on the i transportation network. This demand has been ignored by KLD. For these reasons, the LILCO evacuation time estimates are too
' low.
Given the large number of work-to-home, school-to-home, shopping-to-home, and other trips likely to be generated, there 4 l .
l L l could be as many pre-evacuation trips as evacuation trips. (See Herr Testimony on Contention 65). In fact, considering that many households may have more than one car in use during i the normal working day (i.e., two working spouses, student children, etc.) this estimate may be low -- that is, there could be even more pre-evacuation trips than evacuation trips as families assemble prior to their leaving the area. LILCO states that persons who are not yet evacuating will be permit-ted to travel along the evacuation route network and will be permitted to travel in any direction. (Appendix A at IV-8). Yet, there is no provision for this demand in the trip genera-l tion, distribution or assignment submodels. The KLD model sim-ply does not model pre-evacuation trips. One result of failing to account for pre-evacuation trips is that the trip generation input to the model seriously underestimates the number of trips originating within the EPZ and traveling along the evacuation routes. In short, the demand on the network is underestimated. The failure to include pre-evacuation trips means that the resulting addition-al traffic volume, and its contribution to conflict, conges-t i.o n , and increased evacuation times, has not been measured or I considered in the LILCO evacuation time estimates. l l _ 11 -
. = . .
An aspect of the failure to model work-to-hume trips real-j istically is evident when viewing how the model treats the j large employers within the EPZ, such as Brookhaven National Laboratories ("BNL") and Grumman. The LILCO Plan calls for the direct evacuation of workers from BNL and Grumman. (Appendix A at IV-120; IV-125). Workers from these facilities are sent out of the EPZ directly from work, ignoring the likelihood that i some (or perhaps most) of the employees who reside within the EPZ will want to assemble family units within the EPZ before evacuation. (See Testimony of Drs. Erikson and Johnson on Con-tention 25 (Role Conflict)). Such pre-evacuation trips will add to the aggregate number of trips on the network -- increasing volume and evacuation times. Many other workers (aside from the BNL and Grumman workers ! referenced above), both within and outside of the EPZ, will seek access to the evacuation network, primarily to reunite with family. (See Herr Testimony on Contention 65). Again, the additional trips generated by those workers and the result-ing additicnal demand on the roadway system are not considered in the KLD trip table. Yet another ef fect of the failure to model pre-evacuation trips is thtt KLD has not accounted for potential conflicts in i l l 1 l t- %-mm- ,- - - -
l traffic flow between pre-evacuation and evacuation traffic. As will be discussed later, the LILCO Plan assigns each driver a specific route out of the EPZ which allows for no deviation. l l However, even KLD admits (Appendix A at IV-9) that evacuating traffic following those prescribed routes may well come into conflict with pre-evacuation flows. For instance, pre-evacuation traffic will seek to cross evacuation routes, thus interrupting evacuation flow. Given the fact that there will be a large number of pre-evacuation trips, such interrup-tions could be substantial.
- The potential for conflict is graphically illustrated in Figure 3 to this testimony. Earlier this year, KLD contracted with the National Center for Telephone Research (NCTR) for a survey of commuters residing within the EPZ. Using data from this survey, Figure 3 plots the volume of work-to-home travel (ignoring the many other types of pre-evacuation trips) against the volume of evacuation traffic as reported in Appendix A.
1 Ignoring for the moment that both of these distributions are optimistic (that is, both distributions would be shif ted to the right if they reflected more realistic assumptions of the mobi-lization and evacuation process) it is readily apparent from i~ Figure 3 that during at least the first hour of an evacuation, a substantial portion of the pre-evacdation and evacuation r l l i
FIGURE 3 100 , g 95 -PERCENTAGE OF COMMUTERS RETURNED HOME ............! (Source: EFZ 0-D Survey 90 ~ conducted by NCTR) i 85 80 - l PERCENTAGE OF EVACUEES 75 - DEPARTED (Source: Table X, Appendix A) 70 - 65 - 60 - I g ..: 55 - e :
$ 50 - ****I -
g :* 2 45
=
4 : i , 40 - 35 - a I 30 - i l 25 - : 20 - :"*
- 15 - !
10 -" 5 - I I i t t t i 15 30 45 60 75 90 105 120 Time (minutes) 9 e e
. * . e i
traffic will be attempting to use the evacuation network at the same time. (Compare for example the percentage of evacuation i and pre-evacuation trips at 15 and 45 minutes into the evacua-tion.) Thus, there will be much more demand on the roadway network than modeled by KLD, as well as a significant po t enti al for conflict in traffic flows. This will lead not only to i higher evacuation times, but will lead as well to an inc r e a s. - in the time it will take for people to mobilize themselves and prepare to evacuate, since people attempting to get home or to embark on other pre-evacuation trips will be slowed by conges-tion. Thus, they will be delayed in beginning their evacuation ( trips.
, Similarly, the KLD model has failed to account for any l conflicts between evacuating traffic and " background traf-i fic" -- that is, traffic already on the network when an evacua-tion begins, such as cars or trucks making normal trips on the l network. KLD has properly assumed that the network is not j
empty at the time evacuation begins, but, without articulated i basis, KLD apparently assumes that all of the background traf-fic is moving with the evacuation flow. This is an unrealistic assumption. Traffic on the network is, of course, going to be moving in all different directions, not merely along LILCO's prescribed evacuation routes. Thus, there will be conflict l l [
l 1 1 between evacuation traffic and background traffic resulting in even further delays for evacuating traffic. In failing to l f include this conflict in its modeling effort, KLD has reduced 1 evacuation times -- but at the expense of accuracy, reliability i
- and reality.
i i Q. Can the pre-evacuation trips you have discussed be modeled realistically? I , A. Yes. The work-to-home, school-to-home and other such i traf fic could be modeled with existing technology. In fact, i conventional traffic planning models are usually applied for + this sort of purpose. Inputs would include data such as school 1 < bus schedules, commuting times under normal conditions, and a probabilistic distribution of trips to and from schools, em-ployment centers, shopping areas, etc. For example, the Voorhees Urban System Model mentioned previously predicts work-to-home, home-to-shopping, work-to-shopping, and returns in an , urban context under " normal" conditions. Such a model might be included in the KLD procedures to provide a more complete trip ! table. This would provide a more complete trip table. Obvi-1 ously, it would add to the aggregate number of trips modeled and present the planners with counter-evacuation traffic flows. It would also influence the zone specific trip generation time 1
- - a-->- , - - -,e + . - - . r, - . - - + ,
.= ..
I 1 patterns. Thus, this more realistic trip table would produce ) l more congestion and substantially increased evacuation times. i Q. If KLD has not modeled pre-evacuation trips, how has it determined trip frequencies -- that is, the time over which evacuation trips will be generated? A. Trip frequencies for each source node are set forth in Table X of Appendix A. In reviewing these frequencies, I have found that the vast majority of them follow the pattern below:2/ 0 - 15 minutes - 2.5% of trips 15- 30 minutes - 7.5% of trips 30- 60 minutes - 35.0% of trips 60- 90 minutes - 45.0% of trips 90- 105 minutes - 7.5% of trips 105-120 minutes - 2.5% of trips No justification for this pattern is provided in Appendix A, nor has KLD conducted a serious analysis of these frequencies.
~2/ A few source nodes, apparently representing major facilities, load the network more quickly on the assump-tion that people in such facilities can leave the EPZ more quickly. Lieberman Deposition at 64-65. Note, however, the problem discussed above with the assumption that all employees will leave the EPZ directly.
1
1 I While Mr. Lieberman testified during his deposition that he considered the figures in Appendix A to be reasonable based on certain assomptions concerning work-to-home travel (those assumptions are still unclear), he specifically stated that the numbers do not reflect school-to-home trips (or return) (see Lieberman Deposition at 39); nor apparently, do they reflect notification times or any of the other possible types of pre-evacuation trips such as those made for the purpose of gathering family members, provisions, etc. As mentioned above, if such trips were considered, the evacuation times and mobili-zation times would be much longer, i Q. It appears that KLD has also adopted a 20-minute "mobili-zation" period which represents the period from the declaration of an evacuation to the time the first car leaves home. Do you have an opinion on this portion of the time estimates? A. The 20-minute " mobilization" period is tacked on to the beginning of KLD's time estimates, but I believe its derivation is not based on valid assumptionc or data. In response to interrogatories, LILCO has stated that the 20-minute mobiliza-tion time period is based on a survey of commuters conducted ( for KLD by the National Center for Telephone Research. In this survey the' question was asked: "Approximately how long does it l l l l
usually take you to travel home from work?" The results showed a median commuting time of about 20 minutes. Apparently based on this response, KLD adopted a 20-minute " mobilization" l period. However, there are clear flaws in usin3 these data for l that purpose. I First, there is no explanation or logic to basing the time between an order to evacuate and the departure of the first evacuation trip solely on median commuting time. If work-to-home pre-evacuation trips are to be the basis for estimating the period of time between a recommendation to evacuate and the generation of the first evacuation trip, more than just commuting time must be considered. Second, the survey results are of questionable applicabil-i ity for an additional reason. The response to the question asked appears to include only driving time. Thus, it does not reflect the time that will exist between the order to evacuate l and a commuter's awareness of that order; nor does it take into account the time it takes to secure the work place (close and lock cash registers, shut down machinery, etc.), call home, walk to the car and exit the parking lot. (See Herr Testimony on Contention 65). If such factors were taken into account, the so-called 20-minute " mobilization time" would probably expand significantly.
.* .e Third, as discussed above, there will be many other types of pre-evacuation trips. Most notably, school-to-home travel may take a significantly longer time than commuter travel.
(See Suffolk County testimony on Contentions 68 and 69). Therefore, basing the 20-minute mobilization time on commuter travel only does not reflect the longer periods of time neces-sary for other types of pre-evacuation travel. Contention 65.C.4 Q. What is the concern expressed in Contention 65.C.4? A. The concern expressed in Contention 65.C.4 arises from the destination assignments and evacuation routing strategy adopted by LILCO and reflected in KLD's evacuation time estimates. LILCO has assigned specific evacuation routes and destinations to the households in each of.the 19 zones within the EPZ. The prescribed routes are described at pages IV-87 through IV-178 of Appendix A, Rev. 1. KLD's time estimates assume strict com-pliance with the routes prescribed in Appendix A and strict ad-herence to those routes, without deviation, is considered essential by LILCO to the success of the evacuation: All generated trips must travel from the respective origins (e.g., a home or apartment dwelling) along local, and possibly, " collector" streets to gain access to a link of the evacua-tion network. The precise route taken from the r in- imi ii iiiii . _.m . J
origin to the specified network link is left to , a the discretion of the motorist. It is essential, however, that each motorist enter the evacuation network on the specified links, and on no other network line. If the motorist errs in this respect, he may find it impossible to travel toward his assigned destination without disrupting the flow of evacuating automobiles, increasing his own delay and that of many other evacuees. (Emphasis added) Appendix A, Rev. 1, at IV-19.1/ However, in my opinion, basing time estimates on rigid compliance with the use of prescribed routes is unrealistic and unsound. As Professor Saegert, Professor Herr and the Suffolk County Police Department witnesses have testified, people at-tempting to evacuate from the EPZ during a radiological 3/ This language has been somewhat revised in Revision 2: It is important, however, that each motor-ist enter the evacuation network on the specified links to ensure that he is trav-elling in the proper direction at the out-set of the evacuation trip. If the motor-ist errs in this respect and enters the network travelling in the wrong direction, he may correct his direction or follow the next sign and take another route out of the EPZ. In either case, the motorist is apt to lengthen his travel time relative to executing the proper movement onto the evacuation-network. Nevertheless, the t'ime estimates still assume full'compli-ance with assigned routes. l l m ~ n w
emergency will have their own perceptions of the routes that will present the least risk to them and their families. Their perceptions of the safest way out of the EPZ will be based on their own knowledge of the roadway system existing in the EPZ and will not always coincide with LILCO's strategy. Although LILCO intends to attempt to influence the traffic' flow by use of its so-called traffic guides, barricades, signs, etc. (see Appendix A at IV-5 through IV-79), I concur with the testimony of the duffolk County Police Department, Professor Herr and Professor Saegert that LILCO will be unable to prevent devia-tion from its assigned paths by a substantial number of evacuees. (See Testimony of Suffolk County Police Department witnesses, Philips Herr, and Susan Saegert on Contention 65.) LILCO's strategy of prescribing routes and the failure of the time estimates to consider deviation from the prescribed routes is a major example of the LILCO Plan's failure to account for human behavior. By ignoring deviation by residents or transients within the EPZ, KLD has been able in its model to reduce conflict between flows of traffic and assure that, at least under computer simulation, all available road capacity is put to the best possible use. In doing so, however, KLD models an ideal which has no basis in reality. In other words, the model is not being used to pred'ct what may happen in an l l l
. . . . _ . . - - - - . . . -. -. . -~ .. . - ~ .. _ _ .o ,o i
i i 4 emergency, thus providing realistic information to the l decision-makers; rather it has been used to produce the lowest t l times that a computer could produce. If the time estimates in l Appendix A were to account for deviation (both in terms of the j number or proportion of vehicles deviating and the degree of i spatial deviation), they would be substantially higher. I. Q. Please elaborate further on how deviations from prescribed ] routes will raise evacuation times. j A. The purpose of the use of prescribed routes in the LILCO ! Plan is to avoid conflicts between traffic flows as much as i t possible and to make optimum use of available network capacity. l If people deviate from those precise paths, there will be an i 3, increase in conflicts. Deviant traffic flows will interrupt the flow of traffic along prescribed routes and vice versa. i ) Furthermore, certain routes which many evacuees perceive to be the " safest" will become severely congested, with long queues developing. The result is that there will no longer be an op-timum use of available capacity. While KLD has modeled an ideal evacuation, the time i estimates are not valid because they are not based in reality. The reality is that drivers will take the route out of the EPZ that each perceives to be the safest. (See Testimony of Herr,
- , . . . ~ . . ...,..r . ,, _ . ~ ~ , ,,m, .-. m, -..,m.,.. , , . . _ , - . . . . . _ , , . _ . ~ . , _ . . ..~,_,m-
I .= .. l Saegert and Suffolk County Police Department witnesses on Contention 65). To the extent that people perceive LILCO's prescribed routes to present more risk than an alternate route, substantial numbers will not hesitate to take the alternate route. Many people will deviate unintentionally because they will have forgotten assigned routes or because, as non-residents, they may not be aware of the appropriate route. The Suffolk County Police Department witnesses have de-scribed in their testimony on Contention 65 many prescribed routes and traffic strategies which are counter-intuitive and thus are unlikely to be followed by evacuees. Let me cite an example from Zone B. KLD has devised two routes for this zone. They are: Route 1. Those residents with direct , access to Rt. 25A are to take Rt. 25A east to William Floyd Parkway (CR 46); CR 46 south to the Long Island Expressway (Rt. I-495); Rt. I-495 west to Exit 62, Nicolls Road (CR 97); CR 97 north to the College. Route 2. Those trips originating at, or south of Cooper Street, with direct access to Randall Road, take Randall Road to Whis-key Road and turn west on Whiskey Road. (Access to William Floyd Parkway from Whis-key Road will not be permitted.) Proceed west along Whiskey Road to Ridge Road, then turn south along Ridge Road to Middle Coun-ty Road (Rt. 25). Travel west along Rt. 25 to-Yaphank-Middle County Road (Rt. 21), then south along'CR 21 to Sills Road and to the Long Island Expressway (I-495). Enter i
.' .o I-495 westbound to Exit 62, Nicolls Road (CR 97), then north along CR 97 to the col-lege. Appendix A at IV-91. See Figure 4 hereto for a visual depic-tion of these routes. In this case, neighbors are being routed over very diver-gent routes to the same destination. Those who are designated to follow Route 2 are faced with smaller roads and many more intersections. They are also nearly within sight of the 4-lane William Floyd Parkway but will not be allowed access to it, al-though they will know it is one of the major highways out of the EPZ. A reasonable driver assigned Route 2 might elect (1) to proceed north on Randall Road, east on 25A to the William Floyd Parkway, or (2) south on Randall Road to Whiskey Road, then east to the William Floyd Parkway; or (3) south on Randall Road to Middle County Road, then east to the William Floyd Parkway; or (4) south on Randall Road, then west on Middle County Road. In any of the above cases, he will be deviating from LILCO's prescribed routes. In fact, as Professor Saegert states in her testimony on Contention 65, many drivers will purposely deviate from prescribed routes pre-cisely because they are prescribed -- hoping to avoid the con-gestion on those routes. 24 -
' s D
~~-.%,,
h A y J m 4 ,. . 4 2 . i l!a bl / ; = % ~', . %.i '?g u o w -
& $lN
) is,'v \ j '\ . I y
M M 5
~.
/
,, 3 .
A N!
!se e
- a a I e
- s p~ . n , < ei t n -
~g j h{
om/ a/ f *
~ .
'p w N ' h
_,ft.s &.,,-A
\
/.'
'E #
'\
I
'l +
fo~ .' p 7 g\ D i o ~
' a 1 \ y / /' ' % .
g g
^
$ ,q\ f,~'f: .'4q
~
0
?$ ~ -
_ . r- ,i , m en yf e ., i ' g s. (< i d{,1 j, j/ pgji f-Nb }
.T
.A -1 :
., .o .
s./ .
'I
; il-r he., y & ? \ .
.5 g,- \
l > ?d
\ l g o ;
o ,. il s - ' L$ i a \
'f ~'
a Y -
,?, $' \
,1 \{ <
w
./i, ,r, i
. ! >r ;%
/
~E ' '
? Q \%w/ 1
] '$ 2. <: .
i afw ; i
<xg s .
~
g3 ; 3 s e of:=.....v.; ;y T Basam6 /s-g -
/r. , ,
r 7 \ m
y , 7
. Si . = 4 e is (
- ., O I r ) ,
h f t i
- q f a
%- ~/ Based on the testimony of Professors Herr and Saegert and
>g
/thh Suffolk Cotinty Police Department witnesses and my own views '
'/
of thel. LiLCO Plan, I believe substantial numbers of evacuees
* , s -
2 .
' will. deviate from LILCO's prescribed routes, resulting in con-s x ,
flicting traffic ' flows and congestion that KLD has ignored. 4 i , < a
<s, ' This is alls the more true given LILCO's questionable' ability to
/
l i i ,
;' d,irect or control traffic. (See Suffolk County Police
' Dopartment and Saegert Testimony on Contention 65). If the a .
/ ., < evacuation ' time, es timates 'did not assume full compliance with -
- 1. /'
r.eILCO's prescribed reutes, the estimates would be substantially
/
,l,'.
f.
/ higher. . ,
n
- Q. . Explain hr.;,w the unrealistic asdumptions you have described q
l ' aro 'iricorporated in KLD's model, t,
'.t. .
,,, f A. The failure to account for realistic behavior occurs in two parts or'the modeling process. The first'is in the trip
' distribution component which is the process of ' assigning desti-J,'_
, . J-
- nations lexit nodes) outside the EPZ to the various source I' nodes within the EPZ. The second is in trip assignment. Each 1
- 'y , is diseassed below. ,
W. 1 / (, '.
+
,e I
i
, (
i
/
/ l- t f (
- .. .
- ,4
>-* y l i { , l
< 1
\
, 1 i ,
#.* ,e , ,
.,' ' , ~
! < r. ,
, ,N, 4_ f ., _ _ _ _ _ , , , _ _ _ ,
<n---
f.
, ~ -L.. - - - , , , . . ,- ., , - . . - . , . . _ _ _ _ ,, , , , _ , _ _
Trip Distribution In conventional transportation planning, the trip distri-bution step usually involves a quantitative assessment of the attractiveness of each possible destination and a gross measure of the cost (time) of accessing those destinations. The most common general functional form is known as the " gravity model" particularly the " entropy maximizing" formulation devised by A.G. Wilson ("A Statistical Theory of Spatial Distribution Models", Transportation Research, 1967, pp. 253-269). Typi-cally, such a model is used to reproduce human behavior at first, in calibration, then to forecast behavior for future conditions. In other words, the model is calibrated to " fit" real world behavioral data, gathered from a comparable area and 1 time period. After being calibrated, the model is used to predict future distributions assuming the behavioral elements remain the same. The process distributes trips in a fashion which reflects the actual decisions made by travellers under consistent conditions. However, LILCO and KLD have not attempted to make realis-tic predictions of how drivers will seek to choose destinations in an evacuation from a Shoreham accident. Rather, they have arbitrarily assigned destinations (exit nodes) according to no 26 - 9
~ _ - -. _- - - . -. . _ _ .. .* .s other criteria than the planner's judgment as to which origin /d(stination assignments will reduce evacuation times in the model. (See Lieberman Deposition at 75-76; see also Figure
- 2 hereto). Determining where evacuees wil? go requires much more analysis than KLD has attempted. According to the testi-2 many of Professors Herr and Saegert and the Suffolk County Police Department, it is quite possible, indeed, probable, that people will attempt to exit the EPZ through exit nodes other f than the ones assigned. KLD's failure to recognize this fact results in an invalid trip table which, since it provides cru-cial trip input to the model, renders the time estimates invalid.
I Let me give some examples of how destinations have been assigned arbitrarily in manners which are often counter-intuitive. First, looking at Exhibit 2 of Appendix A (Figure 2 hereto), let us compare origin node 29 to origin node 51. Node 51 is about 2 miles east of node 29. Note that traffic from origin node 29 is sent, with some logic, east to exit node 8011. However, people entering the network from origin node 51, over two miles due east of origin node 29, are sent west to exit node 8006. People seeking the safest way out of the EPZ are unlikely to adhere to this sort of counter-intuitive as-signment of destinations. I l
.* .m
( Another example is the assignment of exit node 8005 to l origin node 18. There are 5 other exit nodes closer (in dis-I tance) than 8005 (8000, 8001, 8002, 8003, and 8004). Likewise, the traffic from origin node 36 is sent to exit 8004 which is i more than twice as far away as the nearest exit node (8002) and much further than two others (8000 and 8001). These are counter-intuitive distributions which are unsubstantiated in KLD's methodology discussions. Under evacuation conditions it seems very unlikely 100% compliance will prevail. In fact, any substantial compliance with these would be surprising. (See generally, Suffolk County Police Department Testimony on Con-tention 65). What is likely is that evacuees will elect to go to the , nearest several exits. The analysis might properly have con-sidered the nearest 3, 4 or 5 exits and distributed the trips according to some probabalistic weighting. This was possible within the KLD model but was not utilized. i It is possible to take human behavior into account in assigning destinations for modeling purposes. Consider, for instance, origin node 24. As you can see, again referring to
! Exhibit 2 of Appendix A (Figure 2 to this testimony), that origin node has been " stratified" in order to send traffic t
.o .o j
i 1 l originating from that point in three different directions (see also Appendix A, Table X). While some of the traffic is sent 3 west to destinations only a couple of miles away (exit nodes ! 8006 and 8007), the rest is sent east, traversing the EPZ to the eastern-most exit node. It is clear that KLD has stratified this origin node to reduce congestion and travel f
! time, rather than to reflect human behavior in selecting desti-j nations. However, the distribution of trips from this origin t node demonstrates that it is possible to assign multiple desti-nations to evacuating traffic. Hence, it would be possible for the model to incorporate behavioral variation in the trip dis-tribution stage. KLD, however, has not done so.
i Finally, it should be noted that zones H and I contain two i j large employers, Brookhaven National Laboratory and Grumman Aerospace Corporation. The employees from these installations are assigned exit nodes (and later routes) directly from their place of work. It is quite likely that most of these people 1 l will return home first (whether that home is within or outside the EPZ) and use an exit node other than the one assigned. Again, the result is an invalid trip table and th'us inaccurate time estimates. I f-
,, - _ . , . _ - _ - + , - . - - - -- --,. --,,....._v,--. . , , - , - . . , , , -
.* .o Traffic Assignment The second place in which KLD has failed to account for differing perceptions of people as to the safest route f rom the EPZ is the traffic assignment portion of its model. As discussed earlier, this portion of the model assigns the traf-fic (from the previous steps) to particular routes, cr paths, through the network. The basis of this assignment is an " equi-librium" model which balances demand (evacuating population) and supply (the evacuation network and the capacity of its roads). "
Equilibrium" is accomplished while minimizing travel time and is done through the use of a technique , the convex-simplex method, to solve a non-linear programming problem (S. Nguyen "An Algorithm For the Traffic Assignment Problem" Transportation Science Vol. 8, (1974) pp. 203-216). The demand for transportation between each pair of zones in a network is a function of the transportation service attributes of the network (such as capacities). At the same time, the service attributes are a function of the level of demand. Equilibrium is reached when these two functional relationships are in balance. The problem with using 'an equilibrium model in an evacua-tion context is that an equilibrium model assumes " normal" 9 >
operating conditions, " normal" driver behavior and a stable network structure. As Manheim has written: From the behavioral perspective . . . we see that we must include in any model of path-choice behavior a description of how users perceive the level of service of a i path. [T]he flow distribution rule should be seen as a component of the demand model . . . [t]his presumes that . . . each traveler has full and accurate information about all the paths available to him and about their characteristics, and that the pattern of network flows is so stable over time that his past experience (such as the times over particular routes no longer used by him) is still valid. (Emphasis added). t
! M.L. Manheim, Fundamentals of Transportation Systems Analysis,
- Volume 1
- Basic Concepts M.I.T. Press, at 473, 490-91 i
I (1979).4/ l 1 4/ For further discussion of the assumptions underlying equi-librium models see: R.B. Dial "A Probabilistic Multipath Traffic Assign-ment Model Which Obviates Path Enameration" Transportation Research, Vol. 5 (1971), pp. 83-111. E.J. Judge " Tests of Assignment Accuracy: An Inter-urban Case Study" Transportation Vol. 3 (1974), pp. 25-44. E.P. Ratcliffe "A Comparison of Drivers' Route Choice Criteria And Those Used in Current Assignment Processes" Traffic ^ Engineering & Control (1972) pp. 526-530.
,( Footnote cont'd next page) l
1
.* .s ;
r l When considering a normally operating transportation system, an equilibrium solution can be expected as part of the operator's daily learning experience. On a stable network, such as would exist in the typical commuting situation, individual drivers adjust to aggregate conditions and produce
- an equilibrium solution over time. This is accomplistc l through experience and lessons learned in daily driving on the same roadway system to the same destination.
To illustrate, most of us who drive to work have different routes for different expected traffic conditions. If I leave work for home at 4:30 p.m., I will take a different route than if I leave at 5:00 p.m. because I perceive the risk of being delayed on one route is greater than another at a different time. In the daily routine, I may become delayed by unexpected congestion. If I get stuck one day, I may change routes or de-parture times the next time. My adjustment helps to relieve the congestion situation encountered the previous day and improves my trip time as well. The equilibrium model assumes (Footnote cont'd from previous page) F. Tagliacozzo and F. Pirzio " Assignment Models and Urban Path Selection Critiera: Results of a Survey of the Behavior of Road Users" Transportation Research Vol. 7, (1973), pp. 313-29. l l 1
.' .s that through this learning process, drivers become aware of the characteristics of the network and thus find the optimum solution over time.
However, we cannot expect the population to experiment and { practice evacuation routes in order to reach such an equilibri-um. The stability required for an equilibrium solution will not ':xist. In an evacuation context, drivers will assess the risks of delay differently from the normal commuting situation. For instance, it is easy to imagine that individuals may elect a more circuitous route which is thought to be less densely travelled. A " secure" 30 or 40 minute route might be perceived to be preferable to a route that would "normally" be 20 minutes but which the driver perceives may end up to be much longer as a result of the congestion caused by evacuation traffic. In short, the " equilibrium" solution included in the traf-fic assignment portion of the KLD model makes assumptions based on " normal" conditions that will not exist during a radiological emergency. W= - g
Contentions 65.C.1 through 65.C.3 Q. What is the concern expressed in Contentions 65.C.1 through 65.C.3? A. Stated briefly, the concern is that LILCO's own traffic control plan will not work effectively. Appendix A proposes to control traffic through the use of LILCO personnel equipped with cones and signs. These traffic guides will be attempting to " discourage" non-prescribed movements. Contentions 65.C.1 through 65.C.3 allege that LILCO's scheme will not work and will in fact serve to increase evacuation times. The "discour-agement" process is likely to cause delay and confusion (C.1), which will be aggravated by the potential for conflict between traffic guides and motorists (C.2). Furthermore, the traffic guides may be directing traffic contrary to established signal lights, thus leading to further delay (C.3). The ineffec-tiveness of LILCO's traffic control scheme and the problems it creates will cause higher evacuation times than are reported in Appendix A of the LILCO Plan. Q. Do you agree with Contentions 65.C.1 through 65.C.5? A. Yes. The Suffolk County Police Department witnesses and Professor Herr have testified that attempts by LILCO to
r .* -
" discourage" deviation by motorists will not always work and will, in fact, result in backups and delays. (See Suffolk 4
County Police Department and Herr Testimony on Contention 65). 1 Furthermore, as Professor Saegert and the Suffolk County Police Department witnesses have testified, there is a very real potential for conflicts between LILCO's traffic guides and the evacuees (See Suffolk County Police Department and Saegert Tes-timony on Contention 65). Again, such conflicts would tend to diminish the effectiveness of LILCO's traffic control scheme. Finally, both the Suffolk County Police Department and Professor Herr agree that attempting to direct traffic will delay rather than expedite traffic flow. (See Suffolk County Police Department and Herr Testimony on Contention 65). I believe that LILCO's evacuation time estimates should be higher because all of the above factors have not been taken into account in KLD's model. Q. Please explain further why the failure of effective traf-fic control, if modeled, would result in higher evacuation time estimates. A. The KLD model assumes that effective traffic controls are _ in operation throughout the network. (Appendix A at III-ll; see also, Appendix A at V-2). This assumption comes into play l 35 - O
- - - - - - + ,- -,
.' .* l l
in particular in determining roadway and intersection capscities. The trip simulation and traffic assignment submodels depend heavily on accurate and realistic assessment of roadway capacities. Traffic flow is restricted by the capacity of a roadway section (link) to transmit traffic or by the capacity of its downstream intersection to discharge traf fic to the next link. The inability of LILCO's traffic control scheme to facilitate traffic movement will result in reduced capacities which will, in turn, reduce flow. The reduction in flow depends upon the extent of the breakdown in traffic control. KLD's model, by i assuming effective traffic control throughout the EPZ, does not account for the decreased capacities and flows which will i result when, as my colleagues have testified, traffic control breaks down. Thus, the evacuation time estimates derived from it do not reflect reality and are too low. Furthermore, the ineffectiveness of LILCO's traffic control scheme could lead to greater deviation from the routes and destinations prescribed in Appendix A. The result will be increased conflicting traffic flows which KLD has not taken into consideration. I discussed the result of such deviations in addressing Contention 65.C.4. i l
, .r .s C. Are you aware of any studies which have attempted to eval-uate the failure of LILCO's traffic control scheme? A. Yes. KLD attempted to conduct such a study. It is reported in a document entitled "Estinated Evacuation Times for the Entire Population Within the Emergency Planning Zone for the Shoreham Nuclear Power Station, Considering the Effects of Uncontrolled Evacuation, Voluntary Evacuation, Inclement a Weather and Accidents" (hereinafter "KLD-TM-77"). C. Do you agree with the results of that study? A. Only to the extent that it showed that decontrolling the network (eliminating manned controls) leads to higher evacua-tion times. KLD-TM-77 at page 13 states that decontrolling the network will increase the time estimates for evacuation of the 10-mile EPZ by 1 hour and 35 minutes. I believe, however, that this figure underestimates the effect of decontrolling the net-work. In simulating " decontrolled" conditions, KLD appears to have removed their own unusual, manned controls, but makes the unrealistic assumption that " normal" intersection controls will be in operation and will apply. That is, that signs and signal lights will effectively control driver cehavior. But Professors Saegert and Herr, and the Suffolk County Police l Department witnesses, have testified that under the conditions l
that will exist during a radiological emergency, many people will not obey the rules of the road, or feel constrained by traffic signs and signals. The breakdown of these controls will further lirit capacity and flow, as well as cause further deviation from prescribed routing. Even ignoring the possibility of intentional disobedience on the part of evacuees, " normal" traffic control is not likely under the congested conditions which will exist in an emergen-cy. For instance, as noted in the Highway Capacity Manual, a standard reference for traffic engineers: With no control, responsibility technically is shared equally by all drivers in making sure that the way is clear before proceed-ing through the intersection. . . . If appreciable cross-street volumes exist (at a two-way stop intersection], they may periodically 'take over' the intersection, even at high through volume levels. . . . Highway Capacity Manual, HRB, Washington, D.C. at 156, 158 (1965). Again, as a result of one flow "taking over" an inter-section, capacities and flow will decrease. Thus, KLD-TM-77, in assuming that the public would obey all signs and signals and follow their prescribed routes, shows that KLD did not adequately model an " uncontrolled" evacuation.
* .a If it had, the difference in the time estimates would be
) substantially greater than the 1 hour and 35 minutes it estimates.
, Contention 65.D 1
i Q. What concern does Contention 65.D raise? A. The concern is that LILCO's evacuation time estimates do not consider the effects of certain factors that could reduce 1 road capacity and thus reduce traffic flow. Such factors include accidents, breakdowns, cars running out of gas, the ab-i sence of shoulders on some evacuation routes, road construction or repair, and the abandonment of vehicles. Indeed, Appendix A at V-2 states explicitly that the estimates do not consider the effects of major vehicle breakdowns on the evacuation routes. However, as Professor Herr, Mr. Polk of PRC Voorhees and the Suffolk County Police Department witnesses have testified, such J incidents are likely to be present during an evacuation. These incidents will decrease capacities and thus increase evacuation times. A. Please elaborate on how these factors would increase evac-uation times. (
/
- 1 A. Mr. Polk of l'RC Voorhees has estimated that a range of up to approximately 400 incidents of accidents, breakdowns and cars running out of gas could occur during an evacuation of the entire EPZ. (See Polk Testimony on Contention 65). Consider a two lane road with two way traffic, one direction being assigned evacuation traffic, the other lane being for pre-evacuation or special bus traffic. An accident in either lane could totally block the section of road when opposing traffic streams confront one another on the remaining lane. Even a link with two lanes in one direction experiencing an accident will result in the overall link capacity being reduced by more than 50%, since the merging of the streams will take time, cause conflict, and reduce the speed with which a vehicle will traverse the link. (See Polk Testimony on Contention 65). Of course, an accident blocking both lanes would stop flow com-pletely, especially if there are no shoulders.
The reduction of flow as a result of an accident would be expected to continue for varying durations depending upon the severity of the accident. The absence of shoulders on some roads (see Suffolk County Police Department Testimony on Con-tention 65) will inhibit the clearing of accidents and break-downs along those roads, thus prolonging evacuation times. Similarly, abandoned cars could cause the loss of the use of a lane or a shoulder. In failing to model the possible constraints on the road-way system caused by such likely events, the KLD evacuation time estimates do not realistically model the circumstances that are likely to prevail during a radiological emergency. Q. Has KLD attempted to model accidents? A. Yes. In KLD-TM-77, some estimates of accident effects are set forth. However, in my opinion, this effort was based on unsupportable assumptions and does not provide reliable data. First, compare the number of accidents estimated by KLD (4) with the number estimated by Mr. Polk (up to 141). KLD's estimate is inherently non-credible and appears to be a result of its reliance on inappropriate accident statistics. While both KLD and PRC Voorhees relied upon national accident rate statistics, KLD did not consider vehicle speeds in its calcula-tion. If it had, its estimates would be more in line with PRC Voorhees' estimates. (See Folk Testimony on Contention 65). A second weakness in KLD-TM-77 is KLD's decision to locate all accidents on road links rather than at intersections. (KLD-TM-77 at 25). During his deposition, Mr. Liebermnn iden-tified intersections as the primary constraint on the network. l (See Lieberman Deposition at 158-59). Indeed, it is common sense that conflict is more likely at intersections than along i ; l
l a straight section of road. Thus, the impact of locating
)
accidents only on links is that delay due to accidents is mini-mized in the computer model. A third weakness is KLD's assumption that the slow speed of evacuating traffic will insure that only minor accidents will occur. (Lieberman Deposition at 214). However, traffic early in the evacuation process, traffic electing deviant routes, and pre-evacuation traffic will not ce so restricted in speed. When traffic streams of very divergent speeds contact or cross one another severe accidents, as well as accidents in general, are much more likely. Finally, it should be noted that neither the KLD not the PRC Voorhees estimates take into account accidents which could occur during pre-evacuation trips. Therefore, even PRC Voorhees' estimates are low. Contention 65.E Q. Contention 65.E alleges that LILCO's evacuation time estimates do not account for the fact that buses, ambulances and other such vehicles will be operating in the EPZ. Is the I contention correct? l
- 42 -
O .3 A. Yes. According to LILCO's Plan, a number of buses and am-bulances will be operating within the EPZ during an evacuation. These vehicles are not restricted to prescribed routes and may, i in fact, cross prescribed traffic flows. Interruptions in flow caused by buses and emergency vehicles crossing evacuation routes are not considered in the model. As I stated with respect to Contention 65.C.4, any interruption of evacuation traffic flow would, of course, raise evacuation times, r Contention 65.F Q. What is the concern stated in Contention 65.F? i A. As Professor Saegert has stated in her testimony on Con-tention 65, evacuees during a radiological emergency will expe-rience stress that will reduce and diminish their driving skills. KLD's model makes no provision for reduced driving skills in that the number of accidents, degree of operator de-viation from the " rules of the road," and degree of route devi-j ation are all inadequately treated. Q. How would reduced driving skills affect evacuation times? A. Evacuation times would increase. To the extent that emer-gency conditions will make it less likely that a driver will be able to remember or follow his prescribed route, greater
* ,a ,
) i deviations from prescribed routes result. I have already described the effects of route deviation in my testimony on l Contention 65.C.4. I i In addition, decreased driver skills may also increase the likelihood of accidents, which, as I have testified, will also i raise evacuation times by increasing congestion. Contention 23.D t Q. What is the issue raised in Contention 23.D? A. The issue in Contention 23.D is that voluntary evacuation, the so-called " shadow effect," will add traffic and congestion to the network and therefore increase entire evacuation times. KLD has not considered the effect of the shadow phenomenon on evacuation times reported in the LILCO Plan. Q. Do you agree with Contention 23.D? A. Yes. Drs. Johnson and Zeigler have testified that in the event of a recommendation to evacuate an area around Shoreham because of a radiological emergency, thousands of persons from outside the recommended evacuation zone will attempt to evacu-ate. (See Johnson and Zeigler Testimony on Contention 23). In some cases, this will result in the addition of thousands'of 44 - l l
I . cars to the roadway network of Long Island. The traffic l generated by the evacuation shadow phenomenon will increase the demand placed upon the evacuation network. Yet, LILCO's evacu-ation time estimates in Appendix A consider only the demand i created by persons whose trips originate in the EPZ. As with
- the pre-evacuation traffic discussed previously, failure to consider this demand presents an unrealistically low demand level on the transportation network. Persons evacuating from the East End will be using many of the same roads used by evacuees from the EPZ. Indeed, many East Enders will travel through the EPZ on their way west. Persons traveling from the densely populated area west of the EPZ will make it more difficult for EPZ residents to leave the EPZ, causing a "spillback" effect. Hence, the evacuation times reported in Appendix A are unrealistically short.
Q. Has KLD made any attempt to evaluate the effect of the shadow phenomenon? A. Yes. In KLD-TM-77, KLD attempted to analyze the effect of the shadow phenomenon, finding that for a 10-mile evacuation the shadow phenomenon would raise evacuation times only slightly. Assuming 25% voluntary evacuation and effective traffic control, the study found that evacuation time would
= .- . -. . . - - . .--- . . . - -. .* .3 rise from 4 hours and 35 minutes to 4 hours and 55 minutes.
- With 50% voluntary evacuation and effective traffic control, I
l the estimate rose to 6 hours and 15 minutes. KLD also at-tempted to review the effect of eliminating the special traffic controls discussed in Appendix A in conjunction with the shadow ! phenomenon. Assuming 25% voluntary evacuation, evacuation time was estimated to be 6 hours and 40 minutes; assuming 50% volun-tary evacuation, evacuation time rose to 7 hours and 15 minutes.5/ Q. Do you agree with KLD's analysis? A. No. First, the methods used in this computer simulation suf fer f rom the same flaws as discussed above with regard to trip generation, trip distribution and trip assignment. Second, I concur with Mr. Polk and Mr. Herr that the study is flawed in several ways. KLD has unjustifiably omitted the travel of voluntary evacuees from the East End over the Sunrise Highway, even though an eight-mile portion of the Sunrise Highway defines the southern boundary of the EPZ. (See Polk l testimony on Contention 65). 5/ The addition of inclement winter weather raised the evacu-ation time in the latter scenario to 9 hours and 45 minutes. , i i
Furthermore, KLD concluded that evacuation traffic from i the East End would not interfere with traffic from the EPZ be-cause the East End traffic would " trail" that from the EPZ. (KLD-TM-77 at 8). This conclusion was based on the assumption j ! that due to the absence of sirens outside of the EPZ "the evac-i uation trip generation process would extend over a two-hour period within the EPZ and over a four-hour period outside the EPZ, where there are no sirens." It is not logical, however, that the lack of sirens would cause a 100% longer period in which evacuation trips would begin. Modern communications such I as telephones, radio and television (which KLD elsewhere as-sumes are working properly) all provide instantaneous linkage and are not bound by the 10-mile perimeter. Without accepting i KLD's two-hour trip generation period discussed earlier, I can-not accept the dramatic difference between the mobilization times assumed for those in the EPZ and those outside the EPZ.- The impact of KLD extended trip generation assumption is that the EPZ evacuation traffic " clears" the evacuation network (EPZ) before the traffic from east of the EPZ enters the EPZ network. Thus, the traffic from the east and the EPZ is, be-cause of KLD's assumption, separated in time and, within the computer model, congestion is minimized. I believe this as-sumption and the implications for evacuation are unrealistic ~.
.* . :=
They present a misleading picture of what would happen in a real evacuation. . Although not as obvious, the same criticism may be made j with respect to the area west of the EPZ. The EPZ traffic, under KLD assumptions, would have " passed by" before the West i network begins to empty in any volume. In reality the two streams will be much more contemporaneous resulting in a greater degree of congestion in the area immediately west of the EPZ. This congestion is very likely to "back up" into the EPZ. Clearly this will result in an extension of the evacua-tion times. O. Do you have any other concerns regarding the effect of the evacuation shadow phenomenon? A. Yes. As I mentioned before, destination and route se-lection of drivers is subject to their perceptions of the l status of the entire transportation network they confront including areas outside the EPZ. Many evacuees will be aware that populations outside of the EPZ will contribute traffic to the many destinations and routes they may be contemplating. This will influence route selection, meaning that destination i and route selection will be even more divergent, resulting in increased conflict and congestion as discussed above. However, l l
,
- v%
l l KLD, in assigning routes and destinations, ignores the fact l that evacuees may be influenced in selecting routes by their l l awareness of congestion due to voluntary evacuation. Again, this puts KLD's trip generation and trip assignment process into question, thus undercutting the reliability of their time ] 1 i estimates. If KLD had included the behavioral impact of the shadow phenomenon on route and destination selection its evacu-ation times for Shoreham would be higher. Conclusion Q. Please summarize your conclusions. A. The evacuation time estimates in Appendix A are based i on flawed assumptions and incomplete data. As a result, those estimates underestimate the time in which people will be able to evacuate from the EPZ. Thus, they are not useful tools for decision-makers in the event of a radiological emergency at Shoreham. J i
, , . -- .,, - -- ,=---- ,
- .-.
- A *-
. C sh i ATTACEENT 1 9 4 s
I 0 .3 ATTACIIIENT 1 VITA Bruce William Pigozzi May 1983 Personal: Birthdate: June 14, 1945 Birthplace: Philadelphia, Pennsylvania Family Status: Married, wife's name Mary Joy, no children Addresses: Department of Geography 821 North Foster Michigan State University Lansing, MI 48912 East Lansing, MI 48824 Phone: (517)485-6780 Phone: (517) 355-4652 Education: i Ph.D Decree, Geography, Indiana University, 1979 I M.A. Degree, Geography, Miami University, 1969 A.B. Degree, Geography, Dartmouth College, 1967 Areas of Competence: Economic and urban geography with an emphasis on regional level analysis, quantitative methods and models of urban ond regional economic systems, transportation analysis and planning, employment structures and forecasting models, economic base models and input-l output analysis. ! Emoloyment Experiences: Teaching: Assistant Professor, Department of Geography, Michigan State
' University. 1979 to present (tenured 1983) .
Instructor, Department of Geography, Michigan State University, 1977 to 1979 Assistant Instructor, Department of Geography, Indiana University, 1972 to 1975 Instructor, Department of Geography, Miami University, 1969 to 1972. Non-Teaching: Consultant for the Public Service Commission of Indiana from June 1, 1976 to September 30, 1977 Provided rail planning and state economic l
)
~C sh Vita: B W Pigozzi 2 impact analyses. Consultant for Sverdrup & Parcel and Associates, I nc .,, , Engineers-Architects-Planners, 800 North Twelfth Blvd., St. Louis, , Missouri from June 1, 1976 to September 30, 1977 Consultant for: railroad and state level economic planning process. Director of State Rail Planning for the Public Service Commission of Indiana, November 1975 to June 1976. Consultant for the Council of State Governments. Lexington Kentucky. Summer 1975 Preparation of computer analyses concerning impacts of railroad abandonment. Assistant Director of State Rail Planning Staff of the Public Service Commission of Indiana; through the Center .for Urban and Regional Analysis, Indiana University, January 1975 to December 1975 Publications:(copies generally available on request)
" Spatial Design for Basic Needs in Eastern Upper Volta" (w i th A Mehretu and R Wittick) forthcoming in Journal of Developino Areas.
"A Tool for investigating Tourism-Related Seasonal Employment". (wi th B Stynes) Journal of Travel Research,Vol 21, No 3 (Winter 1983) p 19-24
" Estimation of input-Output Coefficients by Means of Employment Data",
(wi th R Hinojosa) , Envi ronment and Plannino A. Vol ik, No 11 (November 1982) p 1469-1478. Eneroy and the Adaptation of Human Settlements. Edited by H Koenig and L Sommers, Center for Environmental Quality, Michigan State University, 1980. Contributor for transportation issues.
" Interurban Linkages through Polynomially Constrained Distributed Lags", Geocrhohical Analysis, Vol 12, No 4 (October 1980) p 340-352.
" Inter-Urban Linkages through examination of Urban Unemployment Data",
in Proceedings of the Indiana Academy of Social Sciences, Vol. X (1976) p 73-79 Indiana State Rail Plan: Methodoloay Review, monograph submitted to the Public Service Commission of Indiana, May 24, 1976. Later published as Research Paper Number 2 in the " Rail Planning and Policy Series" of the Center for Urban and Regional Analysis, Indiana University Bloomington, Indiana; series editior, William R Black.
O sb , l Vita: B,W Pigozzi 3 1 l Indiana State 1 9a11 Plan: Final Phas? 2, a release to the public and the Federal Railroad Administration of the Public Service Commission i of Indiana, January 1976. Co-author and Directer of State Rail ( Planning.
"The Spatial-Temporal Structure of Inter-Urban Economic impulses" in Tijdschrift voor Economische en Sociale Geoorafie, Vol 66, No 5 (1975), p 272-276. ,;
Indiana Sta'te Rail Plan Preliminary Phase 2, = (Volumes I and 11) , a release of the State Rail Planning Staff of the Public Service Commission of Indiana, October 1975 Co-author and Assistant Director of State Rail Planning Staff. , Indiana State Rail Plan Phase 1, a submission to the Federai Railroad Administration by the Public Service Commission of the State of Indiana, January 1975 Contributor and Assistant Director of the State Rail Planning Staff. U.S.R.A. Seements in Indiana: State Analysis and Recommendations, (Vo lumes 3 and 4) , a submission to the United States Railway Association, January 1975 Contributor and Assistant Director of the Governor's Rail Study Technical Staff. Pacers Presented:
" Regional input-Output inverse Coefficients Adjusted from National Tables" (wi th R H i noj osa) . Association of American Geographers Annual Meetings, Denver, Colorado April 24, 1983
" Regional Urban Systems: Recent Structural Considerations" lavited paper to Urban issues Seminar Series, MSU Urban Affairs, April 8, 1983
" Application of the URBAN SYSTEMS MODEL to the proposed Grand Rapids Beltway" (wi th R Collard) . East Lakes Division of the Association of American Geographers Annual Mettings Kalamazoo, Michigan; November 1952.
" Stability of the Spatial-Temporal Structure of Interurban Economic impulses" (wi th A Gaete) . Michigan Academy of Science. Arts, and Letters 86th Annual Meetings in Kalamazoo, Michigan; March 1982.
" Locating Health Centers in Eastern Upper Volta" (wi th A Mehretu and R Wittick) . Annual Meetings of the African Studies Association, Bloomington, Indiana; October 1981.
" Estimation of input-Output Coefficients from National Employment Time Series Data"' (wi th R Hinojosa) . Mid-Continent Regional Science Association Annual Meetings, Cincinnati, Ohio; May 1981.
i 1 I l .'
- sC)
Vita: B W Pigozzi 4 "The impulse Transmission Model and the Functioning of an Urban Hierarchy". Association of American Geographers Annual Meetings. Louisville, KT; April 1980.
" Empirical Employment Multipliers and Growth Indexes: Spatial tmplications of two Simple Models". Mid-Continent Regional Science Association Annual Meetings, Lincoln, Nebraska; April 1980. ,
" Spatial Process and the Time Dimension: An Employment Example",
preseentation to the Geography Department Colloquium Series, Michigan State University, February 1980.
" Testing Jeffrey's Inter-urban impulse Transmission Model with Polynomial Distributed Lags".. presentation to East Lakes Division of the Association of American Geographers, East Lansing, Michigan, September 1978.
"The 1973 Railroad Transportation Act and its Effect on Indiana",
invited paper to the Conference of the Indiana Area Development Council, Ball State University, Muncie, Indiana, May 1976.
" Changing Service Levels as an Alternative to Rail Line Abandonment:
The Case of On-Branch Costs". (with W R Black) presentation to the West Lakes Division of the Association of American Geographers, Carbondale, Illinois, November 1975 "Towards a Spatial-Temporal Structure of Economic impulses", presentation to the West Lakes Division of the Association of American Geographers, Muncie, Indiana, October, 1974.
" Optimal Utilization of Miami Valley Hospitals", presentation to the Ohio Academy of Sciences Springfield, Ohio, April 1970.
Public Service and Other Professional Activities: Planning Department, City of Lansing, Michigan. Member of the East Side Transportation Study Committee, November 1982 to April 1983 Planning Department, City of Lansing, Michigan. Member of the North-East Area Master Plan Committee, September 1981 to April 1983 Undergraduate Geography Club, presentation titled " Current Michigan Applications of Economic Geograph'y" June 1982. National Science Foundation, Project Reviewer for the Geography and Regional Science Program, Division of Social Sciences (revi ew completed April 1982) . National Science Foundation, Project Reviewer for the Geography and Regional Science Program, Division of Social Sciences (revi ew completed October 1981) . . 1 l
E
. *' so l i
i Vita: B W Pigozzi 5
\
Department of Transportation State of Michigan. Interpretive and analytical services regarding the Urban Systems Model (Alan M Voorhees
& Associates, Ir,c ); an entropy-based urban transport /landuse impact simulation and forecasting model. (Ref erence : Fred H Sanborn, Manager Social and Economic Studies Section, Michigan DOT) March 1981 and continuing. Anticipated products: research products, student internships, and service to State.
Reviewer for Charles E Merrill, Publishing Co. (Textbook reviews) , current. Reviewer for Growth and Chance: A Journal of Reclonal Developm .,t, (review completed December 1980). National Science Foundation, Project reviewer for the Geography and i Regional Science Program, Division of Social Science (review completed September 1980) . Participant in Massachusetts Institute of Technology / University of Toronto Special Summer Program on " Transportation in Developing l Countries" August 1980. I Genesee County Energy Workshop. Participant, January 1980 through June 1980, Community based energy policy for Flint, Michigan. l National Science Foundation. Project Reviewer for the Geography and Regional Science Program, Division of Scoial Sciences, (r evi ew completed September 1979) . Participant, Faculty Seminar " Problems of Land Tenure", with Prof. Anthony Koo (Economi cs) , 1978. Courses Taucht: (currently) GEO 427 - Quantitative Methods in Geography, senior /first year graduate student level quantitative methods, basic statistics, 1977 to present. GEO 811 - Advanced Quantitative Methods in Geography, senior /first year graduate student level methods course, linear model, multivariate statistics and modelling, 1977 to present. GEO 403 - The American City and its Region, senior /first year graduate student level, networks and system of cities, 1981 to present. GEO 213 - Introduction of Economic Geography, freshman / sophomore level course, 1977 to present. l Seminars, topics vary, Fall, 1982 "Recent Developments in Regional Science Modelling: Regional Economic Structures and input-output Analysis", Fall, 1983 (planned) " Location and Allocation Modelling l l I ( l
5
- o 6
Vita: B W Pigozzi l with Emphasis on Public Facilities Locations". Courses Taucht: (previously) Introduction to Physical Geography (Miami and Indiana Universi ties) Introduction to Cultural Geography (Miami Un i ver s i ty) Geography of North America (Miami Un iver s i ty) Geography of Manufacturing (MSU) Location Theory (MSU) Committees and Miscellaneous Associations: College, Faculty Advisory Council, fall 1981 to present. Core Faculty, Center for Advanced Study of international Development (C ASID) 1981 to present. Geography, Advisory Committee 1981 to present. Geography Admissions and Awards, 1979 thbrough 1981 Chairman summer 1980. Geography, Undergraduate Affairs Committee Chairman 1977 to 1979 Geography, Computer Liason Committee 1977 to 1979 Geography, Departmental Standards Committee, 1978. Geography, Library Committee, 1977-1979 Association of American Geographers. (AAG) member. Quantitative Methods and Mathematical Models Specialty Group, AAG, participant. Urban Geography Specialty Group, AAG participant. Regional Science Association. l l l}}