ML19224A334
| ML19224A334 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 11/30/1978 |
| From: | Eisenhut D Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML19224A328 | List: |
| References | |
| NUDOCS 7812270056 | |
| Download: ML19224A334 (21) | |
Text
f @s3~
/V
- s
&v;.
?f 7
.e
.L-
- ] 'j,
~ ~ ~.. _
Y 2
(~-
g-,
- ~
4 s
UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION BEFORE THE ATCMIC SAFETY AND LICENSING APPEAL BOARD In the Matter of
)
)
METROPOLITAN EDISCN CCMPANY, ET AL.
)
Docket No. 50-320
)
(Three Mile Island Nuclear Station,
)
L'n i t E )
)
TESTIMONY-OF DARRELL G. EISENHU' EVALUATION OF AIRCRAFT CRASH POTENTIAL FOR liUCLEAR POWER PLANTS NOVEMBER 30, 1978 7TDd70o5%
- r.;o
TABLE OF CONTENTS 1.0 atroduction
2.0 Background
3.0 Calculation of P 3.1 Training Operations 3.2 Military Operations 3.3 Crash Density, C 3.4 Number of Operations, N 3.5 Crash Target Area, A 4.0 Resul ts - Crash Frecuency 5.0 Conclusions Appendix A - Calculation of Target Area
"} ' ' 311
1.0 Introduction A considerable amount of testimony has previously been givea in this proceeding regarding the likelihood of an aircraft crashing into the Three Mile Island Unit 2 facility. The Atomic Safety and Licensing Appeals Board decision (ALAB-486, dated July 19, 1978) points out there are a number of areas Warranting additional discussion.
This testimony is meant to explain the Staff's approach or methodology, its criteria, its model, and finally its calculations and conclusions.
This testi-many is subdivided into several separable parts addre! sed in subsequent sections of this testimony. These include:
- Background
- Data Base
- Calculation of P
- Results - Crash Frequency
- Conclusions This will be discussed in general and also specifically with rescect to Three tiile Island 2.
This facility is located in Dauphin County, Pannsylvania, and is located about 2.7 miles from the Harrisburg Inter-natier.al Airport.
It is located off to the side of the extended cencer-line of the runway at an angle of approximately 34.
2.0 Background
2.1 Acceptance Criteria For a number of years the NRC Staff (and earl er the AEC Staff) has been utilizing a probabilistic approach for evaluating potential accidents from hazards or activities which cccur in tne vicinity of a nuclear i ' O,7f plant. The general acceptance criteria for this type of an approach is found in NRC Standard Review Plan (SRP) 2.2.3.
The approach used by the Sta is found in SRP 3.5.1.6.
This approach is basically a " yardstick" or screening approach.
First, SRP 3.5.1.6 refers (see SRP Sec*. ion II.1) to SRP 2.2.2 as guidance for the acceotance criteria.1/
1! SRP 2.2.3 states, in fact:
"The identification of design basis events r,ul ti ng from the presence of hazardous materials nr activities in the vicinity of the plant is acceptable if the de-sign basis events include each postulated type of accident for which a realistic estimate the proba-bility of occurrence of potential exposu)es in excess of the 10 CFR Part 100 Guidelines exyeeds the NRC Staff objective or approximately 10~ per year."...
"In view of the low probability events under consider-ation, the probability of occurrence of the initiating events leading to potential consequences in excess of 10 CFR Part 100 exposure guidelines should be estimated using assumptions that are as realistic as is practi-cable.
In addition, because of the low probability events under consideration, valid statistical data are often not available to permit accurate quantitative calculation of probabilities. Accordingly, a conser-vative calculation showing that the probability of occurrence of potential exposures in excess gf the 10 CFR Part 100 guidelines is approximately 10~ per year is acceptable if when combined with reasonable quali-tative arguments, the reaslistic probability can be shown to be lower."
It is my understanding that, with respect to the issue of appro-priate probability criterion, the Appeal Board has accepted for the purposes of the Three Mile Island case, a criterion that "a facility need not be designed to withstand a crash the probability of which is less than approxi
.ely 10 '" (ALAB-486, 8 NRC 9 at 28).
I do not intend to reopen this question by quoting SRP 2.2.3.
Rather, I wish simply to restate the criteria used by the Staff as succintly as possible, showing its two elements - realistic estimates and conservative estimates.
?&
Second, SRP 3.5.1.6 was developed to be used as a guideline to assist the Staff in determining when aircraf t hazards should be further evalu-ated.
It was thus intended to be a " yardstick" for determining when further detailed evaluation was necessary.
Third, the general approach in SRP 3.5.1.6 (Section III.3) expresses, in equation-form, a method acceptable to the Staff for calculating the likeli god of an aircraft crash at a nuclear facility site located 'eithin
'ive miles of an airport.
This is the rame basic equation form used in this aralysis.
It is discussed 6 210,<.
Fourth, SRP 3.5.1.6 presents in tabular form, a set of crash densities for various types of fatai aircraft crashes (per square mile per aircraf t movement). The SRP states that care should be exercised when choosing values for the parameters in une equation contained therein and notes that "the ma?.ter of interpreting the individual factors may vary on a case-by-case basis becausa of the specific conditions of each case or because of changes in aircraft accident statistics," or, ir. Other words, while tha table of crash densities in SRP 3.5.1.6 may be approcriate for determining "ballpark" values, detailed careful examination must be undertaken to ensure the applicability and appropriateness of those values for a specific application.
An estimate based on these "ballpark" values set forth in 3.5.1.6 is useful as a " screening" tool to determine those general cases in wnich potentiai aircrash impact warrants a more careful review. Outlined below are the elements of such a "more careful review".
S ' &?f t
2.2 Analytical Model The general model used by the NRC Staff *a calculate the likelihood of an aircraft crash into the Three File Island nuclear plant is as follows:
P C
x N
x A
=
(the likelihood (the crash density (the no. of air- (the target per yaar that in tha vicinity craft flying area of safety an aircraft of the TMI plant over TMI that related TMI will crash
- in no. of could crash facilities) into the TMI crashes per into TMI) plant) sq.mi. per MVT)
Each of these tems will be described in detail later including the values assumed for the evaluation of the Three Mile Island plant.
Since there may be various types of aircraft using an airport near a nuclear plant, and since the values of C, N and A may be different for these different types of aircraft, a value of P must be calculated for each major different type of aircraft. Therefore, this equaticn may take the fam:
+
P
+ P
^P Pgeneral small large military total
=
aviation air-carriers air-carriers
,(CxNxA) general
( *"* 'om M xA))g (CxNxA)gg) aviation A_C A-C 2.2.1 The Term "C" The value for "C",
the crash density is derived from actual aircraft crash data that have occurred in the past. Crash data should be collected from appropriate sources as well as the total aircraft arrival and departure actions (mcvements) that generated the crash data.
Since such data is for actual aircraft crashes, it automatically includes a variety D
O'h of flying conditions and circumstances.
For example, it includes the variety of flight paths used for arrivals and departures, for the many airportr, that generated this crash data, a variety of weather conditions, and a variety of terrain conditions near airports.
In using this data it is necessary to assure that the airport of concern has no unusual features of terrain, etc., which require adjustment of the gener-1 data.
A detailed discussion of this term is presented in Section 4.1.
2.2.2 The Tem "N" The Tem "C" determined the likelihood of any one overflying aircraft crashing into a one square mile area surrounding a nuclear site.
Since there are more than one overflying aircraft, the value of "C" must be multiplied by the number of ovacflying aircraft. The term "N" is used to cenote the number of aircraft that fly near a nuclear facility such that there may be a risk of crash into the nuclear facility while, mak-ing a takecif or landing approach at the airport of concern.
This, for the relevant operations,is discussed in more detail in Section 4.2.
2.2.3 The Term "A" The Tem "A" repre:ents the target area which a crashing plane must hit in order to present a hazard to the nuclear facilit)
This tem is multiplied times "C" since "C" was calculated to be the number of crashes per movement into a one square mile target.
Not all crashes into a one square mile area around a nuclear plant pose a hazard to it.
Thus, a more precise target area must be Jetermined.
Such a target area is calculated by evaluating the detailed plant dasign end by assuming a angle of impact for crashing aircraft.
sj gp35 The value of "A" will vary for different sizes and types of aircraft since, for example, certain structures of the nuclear plant may not be vulnerable to the crash of a small aircraft but may be vulnerable to the crash of a larger aircraft, hence, the " target area" for larger aircrait would have to include such structures should they exist. This has the general effect of increasing the target area (i.e., the area of critical structures that a potentially damaging aircraft strike must hit to pose a hazard), for increasingly larger aircraft.
Similarly, the target area is usually smaller for smaller aircraft.
It is discussed ir nore detail in Section 4.3.
- 3. 0 Data Base
- n evaluationg the potential for a damagiag aircraf t crash at a nuclear facility near an airport, an appropriate data base must be sued.
In the Three Mile Island proceeding this is discussed in the testimony of the Staff hied on Novemoer 30, 1978, (Read et. al. ).
4.0 Calculation of P The method of calculating the frequency of occurrence of an aircraf t crash into the Three Mile Island facility or any airport with the same range of activities as those at Harrisburg, which in turn could pose a threat to the health and safety of the public, takes the follnwing ecuation form, where P is the probability of the event occurring per year (more precisely P is the rate of occurrence):
Etotal
= P (scheduled air carriers)
+P (non-scheduled air carriers) u,
. - o... 3 d
il ' *. [
+P (training)
+P (commuters)
+P (military)
+P (general aviaticn)
This can in turn be written P
s (CxNxA) sch. A-C cal
=
+ (Cx!1xA) non-sch A-C
+ (CxNxA) training
+ (CxNxA) conmuters
+ (CxNxA) military
+ (CxtixA) g.a.
As previously discussed in this proceeding, the Three Mile Island nuclear facility is designed to safely sitnstand the impact of a 2CO,CCO-pound aircraft at a speed of about 200 knots.
It is therefore assumed that the facility can withsta d the impact of aircraft weighing less than 200,0C0 pounds. Since all air Taxi Ccamuter and ali Gener31 Aviation movements are with aircraf t weigning less than 210,CCO pounds, they can clearly be celeted as contributors to tne overall likelinood of a damag-ing aircraft strike. This leaves an equation in the form:
P (CxNxA)sch. A-C total
=
' (CxNxA)non-sch. A-C
+ (CxNxA) training
^
military f.lf3 v
4.1 Trainino Coerations For training activities at Harrisburg,
't is not clear that training is so unlikely that it should be entirely oisregarded (See Read et al. testi-mony).
Due to an absence of data, the Staff have not developed a specific crash rate for training activities; but has concluded taat it appears reasonao.. or conservati e to use the non-scheduled off-runway rates for off-runway training accidents. Table 8, Note 3, Testimony of Read et al.
4.2 Militar/ Ooerations Operations with military aircraft for the past few years at the Harris-burg International Airport have been examined by the NRC Staff. Although there nas been a significant change in the nature of the Airport in the last ten years, we dc not anticipate any additional significant changes.
During 1977 there was about 80-85 total operations of military aircraft at the Harrisburg International Airport that weighed over about 200,000 pounds.
These movements consist of three principal types of aircraft -
the C5A, Cl41 and the E4A.
As indicated in Dr. Read's testimony filed November 30, 1978, there have been no relevant aircraft crash events for neavy military aircraft.
l...refore, for this specific component trere is no specific basis upon which to calculate a crash rate precisely and therefore not a crash density distribution. However, in the absence of otrer data, we believe t is reasonable, for the types of operations for military aircraft that may be encountered at Harrisburg, to use the same trash density distri-bution as that for heavy aircraft for U.S. non-scheduled carriers.
pa
._.. n %1 j
c s Heavy Aircraft As discussed above, Threa Mile Island I and 2 have been designed to withstand the impact oi a 200,000 lb. aircraft at approximately EJO knots. Consequently, the airplane crash of concern with respect to assessing a risk of serious damage to the plant are those associated with planes in excess of 200,000 lbs.2_/, or the so-called heavy aircraft.
As can be seen from Table 5, filed November 17, 1978, only 4 occurred off the runway. This is too small a number to derive a meaningful crash density distribution.
Further, using the applicants' number of operations of heavy aircraf t (Applicants' Table 13), of approximately 20 million (a number with which the Staff agrees) the crash rate for heavy aircraft would tend to be less than that for all U.S. Air Carrier Aircraf t.
We prefer to use the higher estimate derived frca all U.S. carriers in deriving both the crash rate and a crash density distribution.
- Thus, our equation is now:
total = all scheduled carriers x ' heavy scheduled
- A
+ Call non-schedule:
heavy non-scheduled and military x A x
2_/ See Discussion below - for discussion of speed.
._,.- o r-n 4
t t>V 4.3 Crash Density C As indicated in Section 2.3.1, "C" danotes the crash density which is derived from actual aircraft crash data.
A different value of "C" must be derived for each different type of aircraft that uses the Harrisburg International I.irport.
From previous sections we have determined that only " scheduled air carriers" and "non-scheduled air carriers" need to be specifically evaluated for this purpose.
d A determination of the value of "C" can take several forms.
One can assumr. a uniform, or equal distribution, for all locations between 0 and 5 miles from a runway - but such an ap;?cach would be conservative in some areas and non-conservative in others.
Similarly, the O to 5 mile region could be sub-divided into smaller area regions and the value of "C" could be assumed uniform, or equal, throughout each individual area.
A more mathematically accurate formulation would be to use more elaborate mathematics techniques to calculate a specific value of "C" as it varies for various distances from the end of the runway (we call the distance "r") and at various angles from the extension of the centerline of tne runway (call the angle, 3 [ theta]). This latter methed can calculate a specific value of "C" at any specific location.
Using the data on crashes, number of aircraft operations set forth in the totals of Table 8 Read, et al. and summarized in Table 9 and the cistribution information contained in Tables 9A and 98, the Officc of Applied Statistics has provided such a calculation for tne point located 2.7 miles from the end of a runway, with a 34 angle to the extended a ' 0Ib1 centerline. This is discussed in th' testimony of Drs. Moore and Abramson.
The values derived are set forth in Table III of their testimony.
Because of variations in tre density function, it t,as been calculated sepacately for landing accicents and for takeoff accidents.
4.4 Aircraft Movements, N Table 20, Testimony of Read, et al. sets forth historical data for past operations of heavy aircraf t at the Harrisburg International Airport.
This information was examined to determine an appropriate value for the term N which is the number of operations from that Airport which could fly over or near the Three Mile Island facility, such that if they should crash, then they might effect the facility.
From Table 20, Read et al., approximately 600 operations of heavy aircraf t occurred in 1977.
The Harrisburg International Airport has one runway designated 13 af ter 130 and 310 This laads to two possible arrival (and departure) directions. The Three Mile Island facility is located south east of the Airport and is located off to the side of the extended centerline of the runway at an angle of approximately 34 Since aircraft generally land and takeoff flying "into the wind" at any given time, aircraft generally arrive from one end of the paved runway strip and generally depart from the other end.
Hence, for 600 cperations of heavy aircraft at the Harrisburg Airport, about 30C operations are over each end of the airport runway. Accordingly, about 300 operations cccur at end of the runway nearest Three Mile Island.
3ecause of pre-vailing wind directions at the airport, about 655 of these operations
,styy
-s J
U ts%
are landing operations and 35% are takeoff operations.
Summarizing:
- 195 takeoffs 195 landings +
HIA
+ 105 landings 105 takeoffs -
- Runway 31 TMI X
Runway 13 -
Of the 300 aircraft movements that approach or depart in the gereral direction of Three Mile Island only a small fraction ictually fly near the nuclear plant in such close proximity to pose a potential hazard to the nuclear facility. Because the principal flight paths of the airport tend to direct aircraft in the direction away from Three Mile Island, less than 1/2 of the total operations would fly in the quadrant in which the facility is located.
In order to ensure that we conser-vatively bound our evaluation we assume that 1/2 of all operations that
! the cuadrant in which Three use the runway directed southeast, fl Mile Island is located.
In sumnary, the number of relevant heavy aircraft operations at the pre-sent time from the Harrisburg International Airport for our evaluation is:
5 Landings:
= 98 operations 10c
= 53 operations Takeoffs:
2
< 353
-1 Since about 40% of the heavy aircraft movements are scheduled and since the remainder are non-scheduled (including training and military).
Our assessment is broken down further to:
Operation at end of Runway Nearest Three Mile Island Scheduled landings:
39 Takeoffs:
21 Non-Scheduled iandings:
59 Takeoffs:
32 4.5 Target Area, A 9
tiistorically, the Staff has utilized a target area of 0.01 mi" per square mile per nuclear unit. This value was derived by considering various aircraft descent angles for both takeoff and landing accidents and considering a slidein area for aircraft crashing in front of the nuclear plant and sliding into the plant.
The Staff has performed a detailed evaluation of the Three Mile Island 2
facility and has deternined that the 0.01 mi target area is conser-vative.
Further, the Staff has evaluated tne facility recogni.-ing that it is designed to withstand the impact of a 200,000 pound aircraft and therefore ha: a certain amount of protection against aircraft strikes of larger aircraft.
Considering these various aspects, the Staff has determined that the use o
of a 0.01 mi" target area is acceptable.
It has further determined that 2
a target area of 0.0062 mi2 'or landing accidents and 0.0026 mi for
~~ g
? Wh takeoff accidents is more appropriate and is still somewhat conser-vative. Details supporting these values are contained in Appendix A.
Recognizing these c.onservatisms the Staff's evaluation has been developed utilizing a target area equal for takeoff and landing accidents:
2 Landing Accidents:
.0062 mi 2
Takeoff Accidents:
.0026 mi 5.0 Results - Crash Freauency In summary, the input infomation used by the Staff in its calculations for Three Mile Island evaluations are as sumarized below:
Present Relevant Heavy Movements, N Total Scheduled Non-Scheduled Landings 98 39 59 Takeoffs 53 21 32 Crash Tarcet Area, A For.cneduled and non-scheduled activity, the target areas used are:
2 Landings:
0.0062 mi 2
Takeoffs:
0.0026 mi Areal Crash Density, C Scheduled Non-Scheduled
-3 Landings:
2.3 x 10-4.4 x 10 Takeoffs
- 3. 3 x 10-9 2.1 x 10-3 There values simply need to be put into our equation and ccmcuted to yield the value of the likelihood of a heavy ait craft from Harrisburg craching into the Tnree Mile Island facility.
hOQ P
[(CxNxA) landings total
=
takeoffs scheduled
[(CxNxA) landings (CxNxA) takeoffs 3
+
+
non-scheduled Putting in the values yields:
P
[(2.3 x 10 )(39)(.0062) + (3.3 x d(21)(.0026M total
=
[(4.4 x 10-8)(59)(.0062) * (2.1 x 10-8)(32)(.0026)]
+
[0.056 x 10-8 +.018 x 10-8 + 1. 57 x 10-8 + 0.17 x 10-83
=
P
-3 1.8 x 10 /yr total
=
As described earlier, this value is based on the assumption that there are approximately 600 n:ovements of " heavy" aircraft at the Harrisburg Airport.
6.0 Conclusions The Staff has evaluated the likelihood of a heavy aircraft attempting to either land or takeoff und crashing into the Three Mile Island nuclear facility.
Thi3 evaluation was based on last year's (1977) record of operations which showed about 600 movements of such heavy aircraft I
(200,000 pounds).
We can calculate that the number of heavy operations can increase by a factor of about 5 to 6 and still have a crash probability that is no greater than about 1 x 10-7/yr, i.e., the number of heavy operations at
< ~35d Harrisburg cetid increase to about 6 x 600 = 3600 prior to exceeding the criterion of 10-7 This simple extrapolation must be cautiously assumed, however, since it must also be assumed that the breakdown of " heavy" scheduled and non-scheduled activity does not significantly change (it is presently about 40" scheduled and 60", non-scheduled plus other). This is particularly important since the crash density for non-scheduled activity is generally about an order of magnitude larger than for scheduled.
In addition, it should be noted that in ac;ordance with our Standard Review Plan approach, an attempt has been made to calculate a realistic value for a damaging aircraft strike at the Three Mile Island plant. Although we have attempted to perform a realistic evaluation, we believe that our overall result is conservative for a number of reasons. A major conser-vatism for the evaluation was the assumption of the fraction of heavy aircraft movements that fly in the general direction of the nuclear plant.
Whereas our evaluerion assumed that 50';' of the flights using the Three Mile Island end of the runway fly over the nuclear plant, the actual value will be much less since the flight patn approach directs most i
aircraft in the opposite direction of the extended runway centerline.
It should be notea that although this evaluation was performed Epecifi-cally for Three Mile Island Unit 2, the same evaluation is also applicable for Unit 1.
~ 9 ' e3[8
APPENDIX A CALCULATION OF TARGET AREA FOR AIRCRAFT IMPACT AT THREE MILE ISLAND 2 The following buildings were considered as safety related:
Containment Fuel Handlin, Auxiliary Building Service Building Control Building River Water Pump Building Because of the structures capability to withstand the impact of aircraft of up to 200,000 lb it has been assumed that only head-on impacts by larger aircraft would result in significant damage.
For the Three Mile Island Unit No. 2 aircraft impact analysis, the Unit adjoins Unit No. 1 in the north, is protected by cooling towers in the south, is shielded by substation equipment on the east and is protected by the plant dike system and small buildings on the west side.
Inasmuch as the transit of a large 200,0C0 to 3C0,C00 lb aircraft i
through such structures will crevent the plane frcm maintaining the opti-mum or near-optimum hea:-on impact orientation required for penetration a slide-in area need not be included in the calculaticn of the target area.
Likewise, the calculation of target area can conservatively neglect, because of penetration cacability reduction, glancing impact, aircraft wing impact, and secarate engine impact. The capacity of plant
-r- - n e.
[
C t,
structures to absorb such leadings is estimated to be well in excess of the loads imposed by such situations. Therefore, the analysis has not included wing extension shadow area and has reduced t*-
ontain-ment shadow area from a ninety degree to a sixty degree sector.
In the calculation of the target areas it has been assumed that there is a uniform distribution of incoming planes from the North, South, West and East directions. However, only about 2/3 of the aircraft approaching from the East will reach the plant because of interference with the cooling towers. Similiarly, only about 1/3 of the aircraft coming from the South will be able to reach the plant.
Credit has been given for the shielding of portions of Unit 2 by some of the structures of Unit 1.
No credit has been given for shielding from other structures in the calculation of the target area for the River Water Pumo Building.
2 Aircraft have been assumed to approach equally from all directions.
RESULTS 2
The average plant target area decreases from 0.011 mi at an angle of 5 to 0.0026 at 45.
The attached table sets forth various target areas for arrival and departure accidents. This evaluation clearly deronstrates that the use a:
ne
/
(o
b*
h g
?
2 of a target area of 0.01 mi is conservative.
It also demonstrates that if a more realistic descent angle for arrival, incoming, aircraft of approximately 10 is chosen, the plant target area decreases to 2
U 0.0063 mi. Similiarly, if a more realistic angle of 45 is chosen 2
for takeoff crashes, the takeoff target area decreases to 0.0025 mi,
0 The staff believes that these values of 0.0063 and 0.0025 mi" are conservative because the utilization of such values assume that all crashes will occur with these descent angles whereas certainly some crashes will occur at steeper angles and hence result in smaller target areas.
s Regardless of these conservatisms, the staff's initial evaluation 2
b-l-assumes target areas of 0.01 mi for all crashes.
4 h*
M,+
g
. w.,..,,,
d k (# v
..