ML20079F496

Supplemental Direct Testimony of Bw Pigozzi on Contention 65 Re Evacuation Time Estimates

ML20079F496
Person / Time
Site:	Shoreham File:Long Island Lighting Company icon.png
Issue date:	01/16/1984
From:	Pigozzi B SUFFOLK COUNTY, NY
To:
Shared Package
ML20079F461	List: ML20079F458 ML20079F480 ML20079F489 ML20079F491 ML20079F496 ML20079F500
References
ISSUANCES-OL-3, NUDOCS 8401180284
Download: ML20079F496 (14)

Text

. .

DOCKETED USNRC UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION '84 jajf l7 p Before the Atomic Sefety and Licensino ' Board'

)

In the Matter of )

)

LONG ISLAND LIGHTING COMPANY ) Docket No. 50-322-OL-3

)

(Shoreham Nuclear Power Plant, )

Unit 1) )

)

SUPPLEMENTAL DIRECT TESTIMONY OF PROFESSOR BRUCE WILLIAM PIGOZZI ON BEHALF OF SUFFOLK COUNTY REGARDING CONTENTION 65 (EVACUATION TIME ESTIMATES)

O. Please state your name, occupation and qualifica'tions.

A. My name is Bruce William Pigozzi. I am an Assistant Professor of Geography at Michigan State University, East Lansing, Michigan. My qualifications are set forth in my direct testimony on Contentions 65 and 23.D, filed on November 18, 1983.

O. What is the purpose of this supplemental testimony?

A. The purpose of this supplemental testimony is to address two studies included in the testimony of Edward B. Lieberman, of KLD Associates, which was filed on behalf of LILCO on November 18, 1983. The first study, " Development of Time 8401180284 840116 PDR ADOCK 05000322 T PDR

_q l

l Distributions for Evacuation Events and Activities,"

I (hereinafter referred to as KLD TM-139) seeks to respond to some of the issues raised in Contention 65.A which questions, among other things, KLD's trip generation assumptions. (See my testimony of November 18, 1983 at 9-19). The second study,

" Determination of Varying Route Compliance Levels and of a Pro-posed New Roadway on Evacuation Travel Times Within the Shoreham EPZ," (hereinafter KLD TM-140) attempts to address some of the concerns set forth in Contention 65.C.4, which alleges that people will not follow LILCO's prescribed routes, thus raising evacuation times.

By an order issued on December 23, 1983,1/ the Board ruled that Suffolk County was entitled to depose Mr. Lieberman regarding KLD TM-139 and KLD TM-140. This supplemental testi-mony i s based upon my review of those studies and the facts discovered during Mr. Lieberman's deposition.

KLD TM-139 0 What is your opinion regarding KLD TM-139?

1/ Order Granting In Part Suffolk County's Motion For Discov-ery And Response To Lieberman Testimony And To Revision 2 Of The LILCO Plan (December 23, 1983).

2_

A. KLD TM-139 purports to justify the 2 hour2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> and 20 minute mobilization and trip generation time set out in Appendix A of the LILCO Transition Plan. (See Appendix A at V-7). It does so by establishing time distributions for several of the activities which must be undertaken before the evacuation trip can commence (for instance, the time necessary for the popula-tion to be alerted to an emergency, the time necessary to pre-pare to leave work, the time necessary to travel from work to home, etc.).

I have two major reservations about the study. First, the time distributions are based upon the manipulation of a few basic distributions. The component distributions are Distribu-tion 1, (the alert-notification process, page 3), Distribution 2 (preparing to leave work, page 4), an unlabeled distribution at the top of page 6 concerning the process for preparing to leave home, and Distribution 5 (school to home travel, page 8).

It is important to note that all subsequent distributions in KLD TM-139 are based upon these primary distributions. There-fore, any problems with these primary distributions are carried over into the final distribution (Distribution E), which represents the range of time over which families begin the evacuation trip. The basic component distributions appear to be the product of unsupported and guestionable judgments.

3-l 1

l 1

, - ,n 4---7

1 l

l Those judgments are not specified in KLD TM-139; based on the l fruits of the County's discovery, however, they appear to be based on little or no empirical data. Such a defect in the component distributions casts doubt on the validity of the sub-secuent and final distributions.

My second major reservation about KLD TM-139 concerns its misrepresentation of the relationship between FLD's time distribution curves and those of PRC Voorhees. KLD TM-139 implies that its time distribution curves are in close agree-ment with PRC Voorhees' curves when, in fact, that is not true.

(See KLD TM-139 at 15, 17 and 18).

O. Please explain in greater detail your first reservation which questions the primary distributions upon which the calcu-lation of the final distribution (Distribution E) is based.

A. Distribution ] (KLD TM-139 at 3) shows the distribution, over 5 minute increments, of the alert-notification process --

that is, the time over which the public / 2becomes aware of the

-2/ In deposition, Mr. Lieberman stated that the universe of people involved in the notification process would be "everyone who would be involved in the evacuation process, and that includes the people who live and work within the EPZ and the people who live within the EPZ and work outside." (Lieberman Deposition, January 5, 1983, at 67).

t 1

l

emergency from the moment the sirens sound. According to Distribution 1, 30 percent of the public will be notified with-in the first five minutes after the sirens sound, 50 percent will be notified within 10 minutes, and 65 percent will be notified within 15 minutes. The distribution continues until 100 percent of the public is notified within 40 minutes. There is, however, very little empirical evidence to support this distribution.

In deposition, Mr. Lieberman supported his estimate that 30 percent of the public would be notified in the first 5

~

minutes with only his assumption that everyone within the EPZ will hear the sirens. (Lieberman Deposition at 38). However, I am aware of no evidence suggesting that everyone within the EPZ will hear the sirens. Indeed, there is reason to believe that many people in the EPZ will not hear the sirens because of work or living environments or physical impairment. Further-more, many more may not understand what the sirens mean. Some sirens may also malfunction.

With respect to the rest of the distribution, Mr.

Lieberman admits that it is "largely judgmental, based upon the practice of people in this area to be within the range of a radio at their place of business and having access to information in a timely fashion." (Lieberman Deposition at 38). Yet, there are certainly many homes, offices and businesses where radios are not played. It is simply unreason-able to assume that everyone will be listening to the radio during working hours. However, that appears to be the judgment upon which Distribution 1 is based. If Mr. Lieberman had used sound assumptions in Distribution 1 it is likely that it would have shown a longer notification distribution.

Distribution 2 (KLD TM-139 at 4) shows the distribution, over time, for people within and beyond the EPZ to prepare to leave work. Again, there are no supporting empirical data offered to support these assumptions, and, as even Mr.

Lieberman stated, none are available. (Lieberman Deposition at 42). Thus, there is no way to determine whether Distribution 2 is correct or incorrect.

It is also apparent from Mr. Lieberman's deposition that the time distribution representing preparations to leave home (KLD TM-139 at 4) is speculative and completely lacking in em-pirical basis. Rather, Mr. Lieberman stated that, in developing that distribution, he considered "the kinds of l activities associated with leaving your home for a weekend."

On this basis alone, KLD TM-139 suggests that 50 percent of the 6-l

evacuating population will be ready after 35 minutes, 70 percent after 45 minutes, and 100 percent after I hour.

Perhaps those used to weekend holiday trips or short business trips can prepare so quickly. However, it is likely that a significant proportion of the population may take longer.

Thus, this distribution is in error, resulting in shorter trip generation times.

Distribution 5 represents school to home travel under the

" worst case" which, as described in KLD TM-139, is the situa-tion where an evacuation is required on a school day between the hours of 9:30 a.m. and 1:30 p.m., when buses are at their depots and bus drivers are off-duty. KLD TM-139 then assumes, for Distribution 5, that it will take between 45 and 90 minutes to assemble bus drivers and move the buses to the various schools, and another 90 minutes to transport all of the chil-dren home. The basis for these assumptions is not clear from the face of KLD TM-139; however, Mr. Lieberman stated in his de' position that they were based on inquiries made by LILCO em-ployees.3/ It is clear that this distribution not only does 3/ Mr. Lieberman was unable to state with certainty which persons or organizations LILCO contacted to obtain this information; nor did he verify the information himself, other than to inquire from the school districts whether they transport all of their children by bus or only some children. (Lieberman Deposition at 51-52).

I l

1 not reflect emergency conditions, but it does not even reflect conditions that exist during a " normal" early dismissal of school children, as described by the School Administration witnesses,in their testimony on Contention 25. No provision is made in Distribution 5 for bus driver absences, drivers being first concerned with the safety of their own families, traffic congestion or other prcblems related to an early dismissal dur-ing a Shoreham emergency. (See Testimony of Petrilak, Muto and Smith, and Jeffers and Rossi, and Campo on Contentions 25.C and 25.D). Under emergency conditions, it is quite likely that such conditions may prevail resulting in much longer school-to-home travel times than Distribution 5 assumes. Thus, Distribu-tion 5 is, apparently, without factual basis, and is inaccu-rate.

In sum, it Iz spparent that the distributions in KLD TM-139 are based merely on Mr. Lieberman's assumptions with little or no empirical basis. While the methodology used is basically acceptable, its application to mere assumptions is unsound. Thus, KLD TM-129 provides no support for the trip generation times set forth in Appendix A.

O. You also mentioned a concern the comparisons made in KLD TM-139 between KLD's time distribution curves and PRC Voorhees' time distribution curves. Please explain that concern further.

A. Fiqures 1, 3 and 4 of KLD TM-119 (at pages 15, 17 and 18) are misleading in the comparisons they draw with PBC Voorhees' time distribution curve. In those figures, PRC Voorhees' time distribution curves are superimposed on KLD's curves.

The way in which the curves are superimposed would imply agreement between them. KLD, however, has not accurately represented PRC Voorhees' curve. PRC Voorhees assumes for modeling purposes that the public's evacuation activities commence when there is an advisory to evacuate -- not when the sirens sound at the Site Alert stage as KLD has representd in Figures 1, 3 and 4.

Thus, the PRC Voorhees' curves should be shifted 25 minutes to the right, which eliminates the inappropriate implication in KLD TM-139 that the PPC Voorhees and KLD time distribution curves agree.

O. Do you have any other concerns regarding KLD TM-139?

A. Yes. In its work to home travel distribution (Distribu-tion 3), KLD has elected to ignore the travel times of those commuters travelling more than 75 minutes from work to home.

This does not seem reasonable. While some of these people may i be train commuters as Mr. Lieberman stated in his deposition (Lieberman Deposition at 45), there is no reason to believe that all such commuters whose families are in the EPZ will not i

i J

even attempt to return home to help their families whether they commute by train or by car. By eliminating the high end of the work to home distribution KLD has unjustifiably achieved a re-duction in the tail ends of the subsequent curves, thus decreasing trip generation times.

i KLD TM-140 s

O. What is your opinion of KLD TM-140?

A. KLD TM-140 purports to address the issue raised in Conten-tion 65.C.4 regarding the effect of non-compliance with LILCO's prescribed routes on LILCO's evacuation time estimates. (See generally my testimony on Contention 65 at 19-33). KLD TM-140 concludes that varying degrees of non-compliance will have lit-tle or no effect on evacuation times; however, in my opinion, that study and the conclusions it has drawn are flawed.

l Q. Please explain how the concept of " compliance" is relevant i

to LILCO's evacuation time estimates.

l l A. As I noted in my previously filed testimony on Contention 65, the LILCO Plan has assioned each household a destination node outside the EPZ and a specific route to that destination.

These specific destinations and routes are justified by KLD on I

{ grounds that they produce, on paper, the lowest possible 1

l

evacuation times for each evacuee. It is likely, however, that for a variety of reasons substantial proportions of the popula-tion will not trave) to their assigned destinations or will travel to them using different routes. Since the time ,

estimates in Appendix A assume 100 percent compliance with those routes, it is clear that they are not accurate and sub-stantially underestimate actual evacuation times. (See gener-ally the Direct Testimony of Pigozzi, Herr, Saegert and the SCPD witnesses for a further explanation of this issue).

O. How then is KLD TM-140 flawed?

A. In two respects. First, the study attempted to treat the issue of travel to alternate destinations by identifying, "for each origin node, one or two alternative destinations which were viable and attractive (in the sense of near proximity and minimum circuity of the associated paths of travel)." (KLD TM-140 at 2). The model was then run assigning the "non-compliant" traffic to the alternate destination or desti-nations identified by KLD. The problem with this approach is that KLD limited itself to only one or two alternative destina-tions when, in fact, for a great many of the origin nodes there are many more than only one or two alternative destinations available which are viable (in the sense that they do not l

- II -

require travel toward the plant), non-circuitous and reasonably close to the evacuee's point of origin.

As just one example, referring to the EPZ evacuation net-work as depicted in the map which is Attachment 4 to Mr.

Lieberman's November 18, 1983 testimony, origin node 6 is lo-cated in the area where Ridge Road and Whiskey Road cross. By Table X of Appendix A (Attachment 8 to Mr. Lieberman's testimo-ny), we know that origin node 6 is assigned to destination node 8006, located on the Long Island Expressway just west of the EPZ. It is easy to see from the map of the EPZ network that there are at least six alternative destinations (8001, 8002, 8003, 8004, 8005 and 8007) which an evacuee might reasonably choose. Yet, KLD has limited its choice to only one or two.

In this manner, it has not truly modeled non-compliance, but rather has only loosened its rigid prescribed routing to a marginal degree. As a result, KLD has not modeled the true range of options open to evacuees. Therefore, it is not surprising that its results showed that KLD's concept of "non-compliance" had little or no effect on evacuation times.

The second flaw in the study concerns the method by which the model represents a motorist's decision to change routes while enroute, either to his assigned destination or another l

l -

12 -

l l

i

one. To account for this behavior, KLD introduced a change in its model called " queue adjustment." The gueue adjustment process operated in the KLD model when a stream of traffic com-ing to an intersection was confronted with total blockage on the downstream link. When such conditions occurred in the model, the model assigned the traffic stream to a different link, thus causing those cars to " deviate" from their assigned link.

There are two problems with is approach. First, it actu-ally occurs very rarely in the KLD model. The KLD model bal-ances traffic flows such that there are very few, if any, total blockages for more than the shortest time (in the computer model, not in reality). Mr. Lieberman described the frequency of this situation occurring in the model as follows: "it is a minority of the time on a minority of the links. I really couldn't ouantify it." (Lieberman Deposition at 33 (emphasis added)).

Second, the element of non-compliance addressed in the KLD model's " queue adjustment" process is only one aspect of non-compliance -- that is, the drivers' reaction to a specific and, in the model, rare traffic condition which is triggered only l upon actual confrontation with traffic blockage. It does not

begin to address the larger notion, as expressed in Contention 65.C.4, of drivers electing non-compliant routes in anticipation of traffic conditions over the network, based upon their advance perceptions and knowledge of the network.

Thus, I find the examination of non-compliance in KLD TM-140 to be inadeouate and misleading in its unsuccessful at-tempt to integrate human behavior into KLD's model.

_ _