ML19263C469

Second Suppl Initial Decision Concluding That Possibility of Accident Involving Liquefied Natural Gas Cloud from Tanker Is So Remote That Units Need Not Be Designed to Protect Against Such Accidents

ML19263C469
Person / Time
Site:	Hope Creek
Issue date:	04/13/1978
From:	Eva Hill, Luton E, Paris O Atomic Safety and Licensing Board Panel
To:
Shared Package
ML19263C468	List: ML19263C467 ML19263C469 ML19263C470 ML19274D768
References
LBP-78-15, NUDOCS 7902260109
Download: ML19263C469 (57)

Text

.

Cite es 7 NRC 642 (1978)

LBP.78-15 SECOND

- SUPPLEMENTAL INITIAL DECISION 7

UNITED STATES OF AMERICA O

NUCLEAR REGUl.ATORY COMMISSION I

Appearances 6

(O ATOMIC SAFETY AND LICENSING BOARD Troy B. Conner Esq., and Richard Fryliig. Jr., Esq CQ for Public Service Electric and Gas Company and (Q Edward Luton. Chairman Atlantic City Electric Company, Applicants O

Ernest E. Hill O

Oscar H. Paris Peter A. Buchsbaum, Esq and Robert Westreich, b j

Esq., for the Joint Intervenors in the Matter of Docket Nos. 50-354 50-355 Richard L Black, Esq., for the Nuclear Regulatory PUBLIC SERVICE ELECTRIC AND Commission Staff GAS COMPANY ATLANTIC CITY ELECTRIC Introduction COMPANY In accordance with ALAB-429 (August 24,1977), this Second Sup.

(Hope Creek Generating Station, plemental Initial Decision reexamines the question of whether the Hope Units 1 and 2)

April 13,1978 Creek Generating Station, Units I and 2 need be designed so as to protect against flammable gas cloud accidents.' Our order of January 26, 1978, Upon reconsidering whether Hope Creek, Units I and 2, need be de-stated our conclusion' on this issue, and promistd a statement of the signed to protect against flammable gas cloud accidents (see ALAB-429,6 reasons for it in the form of an initial decision which would be issued at a NRC 229 (1977), where the Appeal Board reversed and remanded LUP later time. This Secc,nd Supplemental Initial Decision accomp!ishes that 22,5 NRC 694 (1977), on this issu ), the Licensing Board concludes, on the purpose.

basis of probability values it has calculated from figures in the record, that the plant need not be so designed. It further reiterates its earlier conclusion

1. LNG TRAFFIC that the environmentalimpacts of such accidents that might affect the plant are so remote and speculative as to obviate the need for a supplemental en-A. Ships Per Yesr vironmental impact statement.

1. Taking note of differences in the estimates by the parties of the RULES OF PRACTICE: EXPERT WITNESSES number of LNG tankers that would pass the plant each year, the Appeal Board directed the Licensing Board to look again at this man an remand The testimony and opinion of a witness who claims no personal knowledge of or expertise in a particular aspect of the subject matter of his

'New evidence concerning this matter was received by the Board at he rings held on November 3-4, 1977, and January 10. 1978.

testimony will not be accorded the weight given testimony on that question from an expert reporting results of careful and deliberare measurements.

80n the basu of thmidena, we calculate a conservative probablity of 2.4 s 10 per year as 4

the probability that a nammable gas cloud accident could affect the llope Creek plant.

This value is smaller than the t x 104 (for a conservative calculation) probability figure TECIINICAL ISSLES DISCUSSED: Probability of postulated flammable stated in NUREG-75/087, which contemplates 63 regard of accident events having such low gas cloud accidents caused by LNG and LPG tankers, which could affect probabilities of occurrence. We therefore conclude that the plant need not be designed so as t Pr tect asainst flammable sas cloud accidents.

plant; use of regulatory guide (NUREG-75/087).

Order, January 26,1978.

643 642 1

e (ALAB-429, 6 NRC 229 at 235-236). The Applicants assume that 292 All estimates greater than 360 are, we believe, unreasonable. In reaching our tankers would pass the plant each year if the West Deptford LNG terminal decision, therefore, we have relied on the conservative estimate of 360 LNG is approved, constructed, and goes into operation. This figure is derived tankers passing the plant each year.'

from the Draft Environmental Impact Statement (DEIS) prepared for the West Deptford facility by the Federal Power Commission (FPC)(Kalelkar B. Accidents Per Mlle Supplemental Testimony at 4). This estimate of flect size is based on an assumed length of trade route and the daily gas output of the West Dept-

4. The Appeal Board was unable to accept the estimate of collision rate ford terminal (Board Ex. 2, p. 2).

applicable to LNG tankers, which was derived by the Applicants and found

2. The Regulatory Staff intitially estimated that 360 LNG tankers would reasonable by the Staff and the Licensing Board, for a number of reasons pass the plant each year.' This estimate was based on the assumption that (ALAB-429,6 NRC at 236-239). On the remand, the Applicants and the the Raccoon Island as well as the West Deptford terminal would go into Staff independently derived accident rates and provided this Board with the operation, and on the rationalization that traffic and safety constraints on bases for those estimates (Kalelkar Supplemental Testimony at 9-21; Staff the Delaware River would limit LNG traffic to an average of one tanker per Supplemental Testimony at 12-21). We shall now consider these new day, allowing 5 days per year when inclement weather would prohibit estimates and in the cottse of this review address the questions and con-LNG tankers on the river (Read Supplemental Testimony at 6-8). In sup-cerns raised by the Appeal Board.

port of its estimate, Staff offered several alternative methods of estimating

5. Since data on actual LNG traffic on the Delaware is lacking and future LNG traffic. The only one of these which we consider to be of par-oceanic LNG traffic is not representative of conditions on a river waterway, ticular use is the estimate that the West Deptford facility would be served by both Applicants and Staff calculated accident rates using historical data on 390 tankers per year if it operated at 100 percent capacity 100 perant of the large conventional vessels on the Delaware (Kalelkar Supplemental time. The figure 292 used by the FPC and adopted by the Applicamsis based Testimony at 4; Read Supplemental Testimony at 14-17). The Applicants' on the reasonable assumption that the facility would operate at 75 percent accident data were taken from the U.S. Coast Guard's casually reports up-of maximum capacity because of seasonal demand (ibid.). Lieutenant Stan-dated from fiscal. years 1 % 9-1973 to fiscal years 1%9-1975 (Kalelkar Sup-ton, U.S. Coast Guard, testified that while he could not estimate how many plemental Testimony at 5,18-19). The Staff obtained accident data from LNG / LPG transits could be accommodated on the Delaware per year, he the annotated accident narrative of the Captain of t'he Port of Philadelphia, would agree that one every day of the entire year was probably impossible U.S. Coast Guard (Staff Supplemental Testi nony at 14). Both relied on the (Tr. 3442). Because of the cancellation of the Raccoon Island application, U.S. Corps of Engineers' Waterbone Comn crce of the U.S. for their data Staff ultimately concluded that its estimate of 360 tankers per vear was on the traffic at risk (id.i Kaleliar Supplemental Tc'stimony at 8).

overstated, and it adopted the fi,ure of 292 as being more reasonable (Staff

6. It was demonstrated previously, and accepted by the Appeal Board, f

Proposed Findings at 6).*

that the flammability range (**catchmens 6 stance") of an LNG vapor cloud

3. We find the estimate of 292 LNG tankers per year to be reasonable generated by the maximum credible spill volume of one cargo tank is 12 and Staff's original estimate of 360 to be both reasonable and conservative.

nautical miles (ALAB-429,6 NRC at 242). Consequently, only accidents that can occur within a 24-mile section cf the river (i.e.,12 miles on either

'The Appeal Board stated that it ided to understat.d why the Staff used the figure 360 mhen side of the plant) are considered to be relevant to the analysis (Kaleikar Sup-its report stated that LNG traffic would be about 400 tankers per year (ALAI 1429,6 NRC at plemental Testimony at 9-10).

~

2n,. Staff witness Read testified that the figure of 400 was obtained by rounding 360 to one

7. Once the historical data were obtained, they were analyzed for their Yap q%

%m G uqu gMm uh mM&du hm yNk ca f r the si ptford fa ihty is still pending before the Federal he Ten ec Energy Regulatory Commission (FERC), the successor agency to the FPC. An FEIS for West Applicants and Staff agree in their general approach, they differ in the Deptford was expected to be citculated by FERC in January 1978 (Arvedlund Prepared criteria used to screen the casualties in order to determine which are rele-Testimony at 3). In the DEIS, the FPC staff recommended that the application be denied; this vant.

recommendsn may or may not be adopted by the FERC commissioners (Tr. 3368). Staff witness Zerb, from FERC, testified that the " trend" was not to approve applications for 1.NG terminals in populated areas on inland waterways, but he could not say that disapproval

'Our decision rests on probability values calculned by us, using figures from the record which we consider to be reasonable or both reasonable and conservative.

would be the result in every s0ch case (Tr. 3366).

i 644 645

,l

.o 8.' Because of the physical characteristics of this 24-mile section of the the grounding in its data base might not be relevant because the accident river, and the fact that LNG tankers must operate in compliance with an contained conditions that would not be expected on the section of the order of the Captain of the Port of Philadelphia (COTP),* LNG vessels in Delaware River being considered (Read Supplemental Testimony at 19).

this section of the river:

11. Joint Intervenors contend that groundings should not be climinated

• will not be moored; from the data base for several reasons. They allege that there are indeed will not be in an area of industrial docks or piers; hard areas on the river bottom in the 24-mile river section of interest, will not encounter any area of hard or rock river bottom, even if the notably in the vicinity of Pea Patch Island (Tr. 3075-76). These hard spots vessels should go out of the channel; are not rock ledges, however, which could cause penetration of the tanks of will not overtake or be overtaken by other ships; an LNG vessel (Tr. 3076,3480). Commander llenn testified that while it is will not meet other ships at bends; possible that there are uncharted rocks in the main channel of the Delaware will rrt meet oncoming ships at a relative speed of greater than 12 River, their presence is not likely (Tr. 3486, 3494). lie testified further that

=

knots; he could not envision the failure of both the primary and secondary barriers will only transit during daylight hours; of an LNG tank resulting from the grounding of an LNG vessel on an un-a will only transit if visibility is 2 miles or greater; charted rock in the river section of interest (Tr. 3480).

will only transit with tug escort;

12. Intervenors contend also that a grounding accident could distori the will be in continuous communication on two radio channels; and double bottom of an LNG tanker, which in turn might result in distortion

=

will be supervised by a U.S. Coast Guard escort vessel.

of the cryogenic tank's foundation. This distortion of the foundation could produce stress on the cryogenic containment barrier which might be relieved Given the mode of operation of LNG tankers in the river section ofinterest, by failure of the barrier (Fisher Testimony at 21). Commander lienn, on the the Applicants determined that the only accident which could produce an other hand, testifie d that the LNG vessels are designed to withstand such an LNG spill wn a collision (the striking of one vessel by another vessel) be-accident without loss of cargo; consequently, he could not envision a failure tween an LNG tanker and another large vessel (draft of over 18 feet) of both primary and secondary barriers as a result of a grounding in the (Kaleikar Supplemental Testimony at 10,14; Tr. 3103).

river section of interest, even at a spced of 15 knots (Tr. 3476-80).' In-

9. Accidents which result from groundings were excluded from the data tervenors point out that under cross-examination Commander IIenn base by the Applicants because LNG tankers are double bottomed and testified that it was possible for a force that causes failure of the primary would not lose cargo in a grounding within the 24-mile section of river in barrier to also cause failure of the secondary barrier (Intervenors' Proposed question (Kalelkar Supplemental Testimony at 11 and Appendix A). This Findings, paragraph 37): The record makes it clear, however, that the section of the river does not have rocky ledges nur hard protrusions which witness meant "possible" in the sense that almost anyfhing can be con-could penertrate the hull (Tr. 3075-76, 3064-66, 3476-80). Consequently, sidered to be possible (Tr. 3492-94). Finally, we note that Intervenor David Applicants concluded that only soft. bottomed groundings could occur, A. Caccia rejects Commander llenn's " opinion" that grounding on a rock which would result in lifting the tanker slightly, thus dissipating most of the in the river section of interest would not cause cargo loss from an LNG kinetic energy (Kalelkar Supplemental Testimony at 11 and Appendix A).

tanker; Mr. Caccia offers his own opinion, asserting that "a 15-knot Board witness Commander llenn, U.S. Coast Guard, testified that there grounding on rocks could cause a[n] LNG cargo loss" (Caccia Proposed would be no loss of cargo from an LNG vessel running aground in the river Findings at 1).'

section of interest (Tr. 3476-80).

13. The Appeal Board noted that the FPC, in its environmental

10. The Staff's selection criteria to determine relevant accidents included statements for the Raccoon Island and West Deptford facilities, included any casualty which might cause hull damage (Read Supplemental Testimony at 20). Thus, its casualty data base included one grounding inci-

'For a description of primary and sconocy barriers. see Commander llenn's testimony at dent which resulted in hull penetration. Ilowever, Staff acknowledged that Tr. 3455-56, 3467-69. 3478-79.

'Mr. Caccia. whose participation in these proceedings was ordered by the Appeal Board in

'Lt. Stanton. U.S. Coast Guard. testified that the order of the Captain of the Pon would be ALAB-429 (6 NRC at 233), has claimed no experitise in the area of flammabic gas tanker traf-in effect whenever an LNG or LPG vessel transits the <iver (Tr. 3438-3439; also see Appendix B fie. We do not, therefore. accord the weight to his opinion that we do 10 the opinion of Com-to Kaleikar's Supplemental Testimony).

mander llenn.

646 647

l gro' ndings as having the potential for causing a loss of cargo from an LNG the Applicants to climinate groundings from their accident data base.

u tanker (ALAB-429,6 NRC at 240). That Board also expressed concern that

16. Ramming accidents were initially eliminated from the casualty data a grounded LNG vessel might have to of f. load some of its cargo while base by both Applicants and Staff, because they believed that there were no grounded or might become more susceptible to collision as a result of being objects in the river section of interest which could be rammed by an LNG aground (ibid). Staff Witness Arvedlund, one of the authors of the FPC's tanker (Applicants' Proposed Findings, paragraphs28-28A; Staff's Pro-environmental statements for the Raccoon Island and West Deptford LNG posed Findings, paragraphs 24-25)." Intervenor witness Dr. Fisher terminals, indicated that the accident rate developed in those documents testified, however, aa: he had observed concrete based pylons for transmis-was relevant to LNG traffic over the entire reach of the Delaware River, sion towers being constructed on either side of the channel 7 or 8 miles rather than the 24-mile segment of interest here (Arvedlund Testimony at north of the llope Creek site (Tr. 3621-22, 3634). Lieutenant Stanton con-3-4). Consequently the FPC analysis of groundings is not directly com-l firmed this observation and indicated that while the pylons were not on the parable to the one we are considering. With regard to a grounded LNG most recent nautical chart for the area, the maritime public had been vessel off-loading cargo, the cargo tanks of such vessels are capable of con.

notified about them by the Commander of the Third Coast Guard District taining the cargo indefinitely, although periodic venting or burning of in a local Notice to Mariners (Tr. 3686-87). The presence of the pylons will vapor from the tanks might be required (Kaleikar Supplemental Testimony be indicated on future charts (Tr. 3687). Lieutenant Stanton was unable to testify as to the depth of the water at the pylons, but at the Board's request at 48). When vented from tanks in this manner, methane mixes well with g

he agreed that the Coast Guard would provide the Board with this and other air, and the flammable vapor hazard extends only a few hundred Icet (ibid).' Such a grounded LNG tanker would either be floated off with tug relevant information about the pylons (ibid).

assistance at high tide, or the cargo might be off-loaded to another tanker

17. Subsequently, the Board received a letter from Captain K.G.

Wiman, U.S. Coast Guard, Captain of the Port of Philadelphia, giving us or to tank trucks on barges (ibid).

1

14. While a tanker was grounded, the Coast Guard would take ap-the location of the towers and water depth at one of them." Tower 97 is located about 800 feet west of the shipping channel and tower 98 is stout propriate action to minimize the threat under the Port and Waterways Safe-ty Act (Tr. 3083). Be that as it may, however, a possible collision of a mov-800 feet east of the channel. The water depth at tower 98 is 12 feet (mean ing vessel with a grounded tanker is already reflected in the data bases of low water), but at tower 97 the depth was known only to be at least 18 feet.

Applicants and Staff as a " collision" (Staff Proposed Findings at 12).

In a later letter, dated December 19,1977, Captain Wiman informed us that

15. The evidence before us concerning grounding accidents causes us to he had learned that the water depth at tower 97 was 34 feet at mean low water. With this indication that an obstacle exists in the catchment distr.nce find (1) the river bottom in the catchment distance contains no rocks or other objects that the Coast Guard considers to be a hazard to navigation; wh:ch an LNG vessel might, indeed, be able to ram, the Board reconvened (2) should an LNG vessel ground in this section of the river, most of the the bearing to take evidence concerning the probab;lity that an LNG vesel energy would be dissipated in lifting the vessel;(3) if the hull of an LNG would ram tower 97. The evidence relating to this particular ramming acci-ship were distorted in a grounding accident, it would not lead to loss of den'. will be considered irtfra.'8 We continue, now, with our consideration cargo;(4) if an LNG vessel grounded on uncharted rocks in the river section of the other parameters associated with the accident per mile problem.

of interest, even at speeds up to 15 knots, it would not result in cargo loss;

18. Joint Intervenors and Mr. Caccia argue that the catchment zone can-(5) the fact that the FPC included cargo loss in grounding accidents in its not be assumed to be safer than the rest of the river with regard to ram-risk analysis for LNG traffic on the Delaware is not relevant to the analysis mings, because it cannot be assumed that docks or piers will not be con-here;(6) a grounded LNG ship poses no unusual threat of producing a flam-structed within the 24-mile river section of interest-during the life of the mable cloud of methane sufficient to endanger the llope Creek plant; and llope Creek Plant (Intervenors' Proposed Findings, paragraph 48; Caccia (7) the risk of another vessel colliding with a grounded LNG ship has been j

adequately considered. We conclude, therefore, that it was appropriate for l

"A ramming involves a vesset striking a fixed and immovable object such as a dock or pier.

" Att parties were prowided copies of this correspondence between the U.S. Coast Guard and

'We note, however, that the Captain of the Por order, which is in effect when an LNG or the Board.

l_PG ship transits the Detaware River, prohituts venting of vapors and requires that the vessel "The Applicant has considered the scenario of an LNG vessel ramming a pier of the bridge

" demonstrate the ability to contain or consume boil-off vapors for a minimum of 21 days" located 15 miles north of the plant (Applicantf Proposed Findings at 4 t). Because this bridge (Kaleikar Supplemerdal Testimony, Exhibit B p. 4).

is outside the catchment distance, it is not necessary for us to examine this scenario.

648 649 l

Proposed Findings at 1). Both Applicants' witness Dr. Kafelkar, and Staffs the river in qu'estion here, and there are no docks or anchorages there that witness Dr. Read, testified that they knew of nothing that would prohibit might be used by other ships (Kalelkar Supplemental Testimony at 10-12, the construction of port facihties within the 24. mile catchment zone during 15-16, Apperidix B). Joint Intervenors maintain that collisions with an-the lifetime of the plant (Tr. 3052,3121). But Dr. Read testified that a facili-chored vessels should be included because of the possibility that an LNG ty to accommodate vessels having a draft of over 18 feet would be a deep.

ship will lose power within the catchment distance and be forced to anchor water port, that all deepwater ports on the Delaware had been found and (Fisher Prepared Testimony at 25; intervenors' Proposed Findings, already had cities on them, and that to dredge and construct a new port paragraph 46). We believe that such an evet has been taken into account by the Apt cants throught the inclusion of the three incidents in which an-li would be a multibillion dollar developm-nt (Tr. 3234, 3258-59). lie could see no need for new port facilities in the catchment zone (Tr. 3233). No chored or moored ships were struck by a ship that had lost power or evidence was introduced to indicate any likelihood that such port facilities steerage. We also note that should an LNG vessel be forced to anchor dur-ing transit of the Delaware River, it would do so under Coast Guard direc-might t'e developed. In the absence of such evidence, and in view of

+

evidence indicating that such a development is not likely, we reject the In-tion and ccotrol, and the Coast Guard's escort vessel would remain with the tervenors' speculative argument that future docks or piers should be ac-LNG ship to supervise it and to alert other traffic that the LNG vessel was at counted for in the accident analysis.

anchor (Tr. 3449,3461). Commander IIenn testified that the Coast Guard

19. The Applicants and the Staff excluded co!!isions between a large ship anchors LPG vessels in the river at the Port of Philadelphia and said,"we and a tug or barge, because such an accident would not cause a cargo spill have had no problems" (Tr. 3461). The evidence, we find, justifies the ex-from an LNG vessel (Applicants' Proposed Findings at 43; Staff's Pro-ci:~ ion of accidents associated with anchored or docked ships, except to the posed Findings at 15-16). Collisions between a large ship and a tug or barge extent that Applicants have included three such accidents in their data base.

often result in great damage to the smaller vessel but little damage to the

21. Intervenors' witness Dr. Fisher initially testified that the Applicants ship (Kalelkar Supplemental Testimony at I4; Staff Supplemental had excluded accidents in fog because Coast Guard regulations prohibit the Testimony at 17; Tr. 3103). Thus, while the Applicants initially reviewed all entry of an LNG ship into the Delaware River if visibility is less than 2 miles collisions involving at least one vessel havir ;; a draft of over 18 feet, those (Fisher Prepared Testimony at 25; Kalelkar Supplemental Testimony, Ap-accidents in which the smaller vessel was a tug or barge were discounted pendix B). Upon cioss-examination, however, Dr. Fisher admitted that he (Kalelkar Supplemental Testimony at 14). Stafflikewise excluded accidents was uncertain whether accidents had been eliminated because they occurred in which ships collided with barges or tugboats (Read Supplemental in fog (Tr. 3667) Our reading of the record indicates that, in fact, no ac-Testimony at 17), Joint Intervenors say that they do not consider this argu-cidents were excluded by the Applicants because they occurred in fog ment persuasive, but they acknowledge that "it may... be true"; their (Kalelkar Supplemental Testimony at 16-17; Tr. 3969). Dr. Kalelkar own argument on the point begs the question (Intervenors' Proposed Find-testified that he examined several collisions which occurred in fog and ings, paragraph 45). We find that the evidence supports the determination which "could have been discarded as not being applicable to LNG tanker that a collision between an LNG vessel and a tug or barge would not cause accident analysis since USCG regulations forbid the entry of an LNG tanker sufficient damange to an LNG ship to result in a spill. Consequently, it was to the Delaware unless the visibility is at least 2 miles (see Appendix B)"

appropriate to drop such accidents from the data base.

(Kaleikar Supplemental Testimony at 16-17). But these accidents were "ex.

20. Accidents involving anchored or docked vessels, or vessels in the amined in detail" and climinated for one or more of the reasons which we process of anchoring or docking, were excluded by Applicants and Staff have considered, supra, and found appropriate'(ibid). Thus, accidents in because these situations and activities will not be found in the 24-mile river fog have been included in the data base by the Appliccats. The Staff, too, section of interest (Kalelkar Supplemental Testimony at 15-16; Read Sup-did not exclude accidents because they occurred in fog (Read Supplemental plemental Testimony at 17-18). Applicants, however, did include three ac-Testimony at 17-19).

cidents in which a vessel lost power or steerage, or both, and struck an an-

22. The order of the Captain of the Port of Philadelphia relating to chored or moored ship; the rationale for including these incidents was that LNG / LPG traffic provides that LNG and LPG vessels will not start a tran-it is conceivable the an LNG ship might be struck by a vessel which loses sit unless visibility is at least 2 miles. Nevertheless, it was not unreasonable power or steerage (Kaleikar Suppkmental Testimony at 16). As we in-for Applicants and Staff to inch'de in their data bases accidents which oc-dicated, supra, LNG vessels will be underway at all times in the segment of curred in fog because, as Dr. Fisher testified, fog over water is often found 651

in patches; hence, there may be times when LNG vessels will operate in over $10,000 damage; it is Staff's opinion that accidents causing less than restricted visibility (Fisher Prepared Testimony at 25). Lieutenant Stanton

$10,000 damage wouid not cause sufficient vessel damage to result in cargo testified that should an LNG vessel encounter fog, it would proceed to the loss from an LNG vesse!(Read Supplemental Testimony at 17). Forty ac-nearest anchorage or continue its voyage to the terminal; but he said he cidents causing more than $10,000 damage occurred during the 8-year would expect the vessel to continue its voyage rather than anchor and thus period examined by the Staff (ibid.). From 40 casualties Staff climinated create a greater hazard (Tr. 3448) lie said, further, that in the early morn-raccidents involving large ships cohiding with small vessels or barges, ac-ing before an LNG ship is to enter the river, the Coast Guard escort vessel cidents involving ships at anchor or docked, other dockside accidents, and will make a transit down river from Gloucester City to determine whether rammings (Read Supplemental Testimony at 17-18). The rationales for there are changing fog or weather conditions which might affect the transit, these eliminations have been discussed and evaluated in paragraphs 16 and also to check navigation aids and look for hazards in the river (Tr.

through 20, supra. In addition, Staff discarded the following: a collision 3447). If the patrol craft encountered bad visibility, the LNG vessel would with a car ferry, on the ground that there are no ferries within the catchment be required to anchor near Ship John Schoal Light to await better weather zone; a barge which exploded from battery hydrogen ignition, en the conditions (ibid)." Considering these facts, we believe that the likelihood of ground that such an accident is limited to unpowered barges which utilize a fog-caused accident is so remote that it need not have been factored into batteries for running lights; an accident in which deck equipment was lost the probability analysis, it was conservative, therefore, to include in the overboard, on the ground that this event would not endanger the hull; and iata base accidents which occurred in fog.

groundings which did not cause bottom damage sufficient to endanger

23. Based on the foregoing analysis, Applicants concluded that the only water tightness (ibid.). The Staff's criterion for eliminating groundings dif-type of accident which could produce an unignited spill of LNG in the river fered somewhat from that used by the Applicants, and resulted in Staff ac-section of interest was a coilision between an LNG tanker and another large cepting one grounding accident in which the hull was penetrated (Read Sup-vessel (Kaleikar Supplemental Testimony at 10). The U.S. Coast Guard plemental Testimony at 19; Tr. 3201-3203). The Staff testimony cast some marine casualty reports used by Applicants for their accident data base doubt on the relevance of this case, however (tDid.).

identified U collisions on the Delaware River during the period 1 % 9-1975

25. After excluding accidents on the basis of all these criteria, Staff was (id. at 13). Of the 67 collisions,25 were discarded at not relevant because left with five ship colli: ions and the one grounding, which it considered to they involved small vessels or barges (id. at 13-15,17-18); 22 collisions were be possibly capable of c.using an LNG spill (ibid.). These six relevant ac-discarded because they involved vessels which were moored or anchored (id.

cidents involved a-total of 11 ships (ibid.). While the Applicants counted at 15-18); one was discarded because it occurred in an anchorage area (id. at each relevant collision as one accident of interest, Staff counted each large 15); five were discarded because they involved vessels which were docked, vessel involved in a collision as one accident of interest; thus Staff obtained docking, or undocking (id. at 17-18); and four were discarded because they 11 accidents of interest from its five collisions and one grounding (Read were found to have occurred outside the Delaware River (id, at 13). Thus, Supplemental Testimony at 15,19-21; Tr. 3123).

the Applicants' examination of all 67 collisions involving any type of vessel

26. Staff counted each vessel involved in a collision as one accident revealed that ten could be considered as capable of causing an LNG tanker because it considered both the struck vessel and the striking vessel to be at spill (id. at 18)."

risk with respect to being a potential source of an LNG spill (Tr. 312). Ap-

24. The Staff utilized accident narratives of Captain Goodwin for the plicants, on the other, considered only the struck vessel to be at risk in this period 1967-1974 for developing its casualty data base (Read Supplemental sense (Tr. 3699-3700). According to Applicants' witness, Dr. Kaleikar, Testimony at 16-17). These narratives are limited to casualties which involved should an LNG ship be the striking vessel, the probability of a spill would

,,Ihe anchorage for i NG ships near Ship John Shoal Light is outside the catchment be zero, or close to it, because the bow of the ship would absorb the energy distance, apparently having been located there by the Coast Guard in 1975 so as to specifically of the collision without damage to the forward LNG tank (ibid.)."

avoid anchoring i NG vessels within 12 miles of the Salem Nuclear Generating Station tTr.

27. The Intervenors' argue that the ramber of relevant accidents are 3460).1he liope Cseck site is immediaiety adjacent to and north of the Satem plant.

underestimated by Applicants and Staff because some of the accidents PI " ",'d 'jd; ",' f'd["[p' ",

" '

• fh Ic 99 which were excluded from the analyses should have been retained in the sis o r

( A1.AB 429,6 WRC at 236h That reduction was based on the same rationale as used in analyz-

.. Based on our study of Board lim. 2 (see p.173). the number one cargo tank on a 125.otx)-

inn the daia for 19691975 (Kaleikar Supplemental Testimony at 18-19).

cubic-metcr LNG carrier is more than 10u reet from the bow.

652 653

data bases (Intervenors' Proposed Findings, paragraphs 35-36). Joint In-we have relied on the Applicants' estimate of number of accidents per tervenors' witness, Dr. Fisher, testified that not only should all collisions be year."

included in the data base, but some of the grounding accidents should be in-

31. As noted, supra, Applicants obtained their estimates of traffic at cluded as well; in his opinion collisions and groundings may arise from risk, or exposure data, from the historical record contained in the U.S.

judgmental errors, misinterpretations of information, or misapplication of Corps of Engineers' Waterbone Commerce of the U.S. (Kalelkar Sup-data to which LNG tankers are susceptible (Fisher Prepared Testimory at piemental Testimony at 8; Tr. 3003-07,3690-93). To obtain the number of 24, 26-27). It is clear, however, that accidentr resulting from errors in judg-ship transits on the Delaware River, Applicants used information contained in a table entitled " Consolidated Report Upbound and Downbound" (Tr.

ment were in no way climinated from the data base. Judgmental errors are an inherent cause of accidents contained in the historical record from which 3691-92). Average miles traveled was obtained from another part of the Applicants and Staff derived their data. We must reject, therefore Dr.

report which gave total tons of traffic moved upbound and downbound and total ton-miles for the same cargo (Tr. 3692). From these dash. with an ad-Fisher's suggestion in this regard. Additional reasons which the Intervenors advanced for not discarding certain types of accidents from the data base justment to account for internal movements on the Delawae, Applicants have been dealt with by us in paragraphs 11,12,18, and 21, supra.

derived their estimate of total ship-miles transited (Kaleikar Supplemental

28. The criteria used by the Applicants and the Staff for determining the Testiminy at 20-21; Tc. 3692-93). The estimate was an average of 9,474 relevancy of accidents were similar, except for the single grounding accident one-way transits per year, each ir volving a movement cf 100 miles, to give included by the Staff. We have found these criteria to be reasonable. We an annual exposure of 9.47 x 10' ship-miles per year (ibid.). By dividing the also find the difference with respect to groundings to be insignificant, estimate of accident rate,1.43 accidents per year (ten accidents in 7 years),

because the evidence we have received on grounding accidents leads us to by this estimate of exposure, Applicants obtained per-mile accident rate of believe that the grounding accident included in Staff's calculation would 1.51 x 10

• collisions per mile-year (Kale!kar Supplemental Testimony at 21).

32. In order to test the validity of the Applicants' estimate, Staff in-not have produced an LNG spill. The historical records used by Applicants dependently obtained an estimate of exposure of 60 miles per port call, based and Staff for deriving their data, however, were different.

29. Staff used as its historical record the accident narrative of the Cap-on an assumed round trip transit of the river from the anchorage off Smyr-tain of the Port of Philadelphia, which at once climinated from considera-na to the Delaware Memorial Bridges, even though' the five collisions and tion all accidents involving damage amounting to $10,000 or less. Staff one grounding selected by Staff occurred over a wider range of the river believes that an accident causing $10,000 damage or less would not cause an (Read Supplemental Testimony at 20-21; Tr. 3124-26). Port calls were used LNG spill. We concur, and we find Staff's final selection of five relevant as a basis for estimating exposure in response :a a suggestion made by the collisions and one relevant grounding to be reasonable.

Appeal Board, and the 30-mile section of the river from Smyrna to the

30. Applicants used as their historical record the U.S. Coast Guard Delaware Memorial Bridges was used because it contains the 24-mile catch-ment zone and is unlike other sections of the river (ibid.). Staff divided its casualty reports. This record did not eliminate accidents on the basis of damage costs, which accounts, in part, for the fact that Applicants obtained accident rate (based on 11 accidents in 8 years) by this estimate of exposure, a total of 67 accidents for their data base compared to Staff's total of only to obtain a per-mile accident rate of 1.5 x 10+ accidents per mile-year 40 casualties." Thus the Applicants worked from a much more comprehen-(ibid.).

sive data base than did the Staff. From their data base, Applicants selected

33. The Appeal Board questioned the validity of the Applicants' original a total of ten collisions which they determined to be relevant to a possible accident rate of 1.5 x 10 *, which had been fouid to be reasonable by the LNG spill. We find this determination to be reasonable. In addition, Suff and by this Licensing Board, because it was unclear why some types of because of the large data base and the conservatism used by the Applicants

,,y in selecting relevant accidents, we find Applicants determination of risk with regard to a possible LNG spill, whereas Staff counted both ships involved in a colli-number of relevant accidents to be conservative. In reaching our decision, sion. Appheants argue that the bow of the striking ship would absorb the energy of the colli-

+>n without damaging the forward LNG tank (Applicants' Proposed Findings, paragraph 24).

"Another difference between Applicants' and Stafrs historical records is that Applicants' Applicants' position is supported by the SAR report, which states. "to crush an LNG ship's data came from the period 1 % 9-1975, whereas Stafra came from the period 19@l974; this bow sufficiently to sclease LNG would require a speed in excess of 30 knots, a speed greater aho contributes so the difference in number of accidents obtained by Applicants and Staff than the ship's capability. "Ihus, there is no risk of a retcase of LNG due to ramming another ship by an LNG ship" (Board Ex. I,12.3.2.3). We accept this argument.

(Read Supplemental Testimony at 15).

I 655 654

per mile-year is obtained (Kalelkar Supplemental Testimony at 25). This accidents were excluded from the data base, and because of "apparently value corroborates those calculated by Applicants and Staff.

conflicting and larger values" (ALAB-429,6 NRC at 238). Applicants have

35. The SAI report utilizes a theoretical mathematical model which provided the requisite explanations for climinating accidents from the data disregards the physical and geographical features of the river (id. at 26; see base, and we have dealt with these in paragraphs 15 through 23. We now consider the conflicting values obtained in other studies, in question are the Appendix A of the SAI report). It predicts a collision rate of I x 104 colli-DEIS prepared by the FPC for the proposed West Deptford LNG terminal Sions per 66-mile transit (ibid.). On a collision per mile basis, this comes to I.52 x 10+ collisions per mile-year over the 66-mile zone of interest to sal (Board Ex. 2), the risk assessment done by Science Applications (SAI) for (Delaware Bay to Wilmington, which includes the 24-mile catchment zone the proposed Raccoon Island LNG terminal," the analysis of LNG marine f c ncern in this proceeding)(ibid.). This independent estimate is also very transportation by Booz-Allen," and a study by the Oceanographic Institute ci se to the values obtained by the Applicants and the Staff and provides of Washington (OlW).n additional corroboration of those values.

34. The FPC study had different objectives from the analyses performed

36. The Booz-Allen study addresses potential LNG tanker shipments to by the Applicant and the Staff (Arvedlund Prepared Testimony at 3). The several major U.S. ports, including the Delaware River (Kalelkar Sup-purpose of the FPC study was to assess the risks from LNG traffic to the piemental Testimony at 27). The analysis for the Delaware River was con-public residing along the entire tanker route (ibid.). FPC used as its data cerned with risks along th; entire river. The approach used by Booz-Allen base a historical record of only tanker accidents in the Delaware, and was similar to that used by the Applicants, except that accidents involving cateulated accident rates for six zones of the river, some of which had beta LNG vessels at dock were included because the zone of interest included determined to be uniquely hazardous (id, at 4). In addition to being based that region of the river where these ships will dock (id. at 27-28). The acci-on different data, the method used by FPC for screening accidents differed dent rate derived from the Booz-Allen study is 2.62 x 10 * "potentially from those used by the Staff and Applicants, because ships experience dif-ferent risks in different zones of the river; consequently, many accidents used senmis incidents per mile" (id. at 28). This value is not directly comparable to the accident rates obtained by Applicants and Staff. It is larger than the in the FPC study would not be appropriate for the 24-mile catchment estimates of Applicants and Staff because it includes dockside accidents, distance of concern here (id. at 5-6). The FPC study gives an accident rate which we have found can be justifiably excluded in analyses with respect to of 1.0 x 10 5 collisions per mile, which is greater than those estimated by Ap-the llope Creek plant. Thus, we find that the Booz-Allen estimate does not plicants and Staff (Kalelkar Supplemental Testimony at 23) The In-e ntradict the estimates of Applicants and Staff.

tervenors argue that the FPC rate should be applied to the lione Creek risk

37. The OlW study addresses the question of the likelihood of oil spills calculation, on the ground that all accidents should be included in the in several U.S. ports, including the Delaware River, in connection with an analysis (Intervenors' Proposed Findings, paragraphs 52-54)." We have eff rt to assess the risks of oil spillage associated with an offshore already found, however,(see paragraphs 15 through 29, supra) that the ac-petroleum transfer system for the State of Washington (Kaleikar Sup-cident screening carried out by Applicants and Staff was justified. Ihnce, we conclude that the FPC rate of 1.0 x 10 5 accidents per mile is not sp-pJememal Testimony at 29). If the OlW results are used to calculate a colli-si n rate per mile, the value 2.2 x 10 S is obtained (id. at 30). This rate is ap-plicable to the problem before us, and we do not consider it as contradicting plicable to the entire river, not just the 24-mile zone of interest in this pro-the rates calculated by Applicants and Staff. In fact, if the FPC analysis is ceeding. It includes many accidents of the types we have found to be applied correctly to the llope Creek plant site, a rate of 1.5 x 10

• collisions justifiably excluded from consideration (id. at 31). Consequently, we do not

"" R isit Assesunent of LNG Marine Operations for Racoon Island, New Jersey." Science consider the OlW estimate to be contradictory of those obtained by the Ap-Appheations, Inc.,1975.

plicants and the Staff.

"" Analysis of LNG Marme Transportation for the Mantime Administration." hootAllen

38. The Appeal Board calculated a casualty rate of 3 x 10 5 from figures Apphed Research, Inc.,1973.

a"oifshore Petrolcum Transfer System for Washington State-A Feasibihty Study."

contained in Staff Exhibit 1-F but acknowledged that it "apparently in-Cludes incidents of all types (collisions, rammings, and groundings)"

Oceanographie Institute of Washington,1974.

N "k"*" "' " ""*" D' " *" "'i'd da 6 me length with regard to the FPC analysis (ALAB-429,6 NRC at 238 and n. 44). The record now makes clear the types (Fisher Prepared lestimony at 2n28). Ilis testimony was rebutted by Staff witnen Arvedlund of accidents which were excluded and why theY were excluded. We need not (Tr. 3316-17,3339 43,3359 61,3374-75). In our siew, the cifort at rebuttal was successful, and go over that again. Because,t includes many types of irrelevant accidents, a,

we need not reniew that testimony here.

l 657 656

we' find the 3 x 10 5 casualty rate obtainable from Staff Exhibit 1-F (at p. 8)

Staff (Read SupplementalTestimony, Appendix A, p. 4-5. Tr. 3282). While not applicable to the llope Creek analysis.

portions of these other studies were of interest, they were not found to be

39. The Appeal Board also suggested that LNG tanker experience in dkertly applicable to the analysis of concern in this proceeding (id. at 5).

ports and harbors might be assessed as a check on the validity of the Ap-Similarly, Dr. Kaleikar testified that a large amount of literature in addition plicants' accident rate (ALAB-429,6 NRC at 238). The Applicants responded to that cited was reviewed for its relevance to the Hope Creek evaluation by estimating that there have been some 9,400 harbor entries or exits by (Kalelkar Supplemental Testimony at 31). The studies which were found to LNG tankers on a worldwide basis without a collision with any other vessel be generally useful were listed in Table 2 of his testimony (id. at 62-63). On-(Kalelkar Supplemental Testimony at 32). Modeling the occurrence of an ly those cited in paragraphs 34 and 35, however, were directly applicable to LNG tanker collision as a Poisson process gives a collision rate of zero, the Hope Creek risk analysis (id. at 31). From this evidence we conclude since there have been no collisions of LNG tankers while underway, and that Applicants and Staff have reviewed and made appropriate use of permits the calculation of a 95 percent confidence limit of 3 for the available literature which is relevant to the analysis of concern in this pro-unobserved event (id. at 33). This confidence limit leads to a collision rate ceeding.

of 3.2 x 10

• collisions per one-way harbor transit (ibid.). Assuming that the

42. Having considered this evidence and made the foregoing findings, average one-way harbor transit of LNG tankers is 8 miles, Applicants we conclude that the accident rate estimated by the Applicants (1.51 x 10*

calculated a rate of 4.3 x 10 5 collisions per mile-year (ibid.). Expressed in collisions per year-mile) is reasonable. It is corroborated by results obtained other words, one is 95 percent certain that the collision rate for LNG in the FPC study and by the theoretical model developed by SAI. Larger tankers underway in harbors is less Ihan 4.3 x 10 5 (ibid.). If the safety values obtained in other studies or from other calculations do not contradict record for LNG tankers persists, this confidence level will 30 down as more the estimates of Applicants and Staff because they are not comparable to and more tanker expericace is accumulated.

them. Finally, we have found the historical record used by Applicants and

40. The Staff responded to the Appeal Board *5 suggestion with regard to Staff, and the data screening procedures used by them, to be adequate and analyzing LNG cxperience using a somewhat simihr probabilistic analysis, justified, for the reasons given, supra.

but its calculation was based on different parameters (Read Supplemental

43. We prefer the Applicants' estimate over the one calculard by Staff.

Testimony at 21-22). The Staff assumed that ship traffic possesses a natural Applicants' figure is based upon an analysis of 67. accidents, compared to attribute, viz., human error, which leads to observed accidents, cnd that Staff's analysis of 40 accidents. Staff's data base included only accidents in-human error is nearly constant with respect to time (id. at 21). Staff then volving more than $10,000 damage. We have agreed with Staff that it is noted that 100 ship-years of LNG tanker experience have accumulated reasonable to climinate accidents that do not cause more than $10,000 without a cargo-loss accident, and calculated that about 4,000 ship-years of damage, on the ground that such accidents would not cause a cargo spill experience would be needed io eest the likelihood of an accident rate of 1.5 x from an LNG tanker. But no such elimination was made in the data base used 10

• per year-mile (id. at 21-22). Thus, Staff concluded that LNG traffic has by the Applicants. We conclude, therefore, that the Applicants' calculation not been extensive enough to be useful in its analysis (id. at 22). With regard is the more conservative of the two."

to LNG port calls, Staff could only say that enough port calls have occurred

44. Applicants' witness Dr. Kalelkar testified that using historical data to provide evidence that "it is likely that LNG tankers have lower casualty based on conventional tankers and freighters to calculate a collision rate to rates than average shipping" (ibid). Based, as they are, on the observation be applied to LNG tankers is conservative, because LNG tankers have colli-of zero accidents involving LNG tankers underway in harbors, we do not s on avoidance systems, favorable horsepower to tonnage ratios, and bow find these estimaies derived from LNG experience to be useful to our thrusters, and also will transit the river with tug and U.S. Coast Guard deliberations. We believe that they are essentially meaningless and that they escorts under rules of the Captain of the Port of Philadelphia for LNG do not in any way challenge the estimates obtamed by Applicants and Staff tankers on the Delaware River; these factors will make LNG tankers less from historical data on conventional large ships.

coll sinn prone than other large ships (Kaleikar Supplemental Testimony at

41. Finally, the Appeal Board suggested that still other LNG studies might coatain information valuable Io the analyses of Applicants and Staff

,,obviously there is no statistically significan: dirrerence between the Applicants' estimate (ALAB-429, 6 NRC at 245, n. 94). Dr. Read testified that many studies and that calculated by the starr. Since the scari's approach =as independent of and somewhat other than those listed in paragraph 33, supra, had been examined by the different from the Applecand, m.ffS c>simate serves to corroborate Applicants' estimate.

658 659

t grounding accidents. Be that as it may, we have determined previously that 19-21). The Staff has taken a similar podtion (Staff's Proposed Findings, grounding accidents by LNG vessels can be disregarded in the risk analysis paragraphs 55,59).

45. Joint Imervenors and Intervenor Caccia, on the other hand, main.

because they would not cause a cargo spill.

46. Second, Dr. Fisher testified that loss of propulsion by an LNG vessel tain that generalized claims for conservatism with regard to LNG vessel would create a hazardous situation (Fisher Supplemental Testimony at 14).

characteristics and operations are invalid (Intervenors' Proposed Findings, Most loss of propulsion accidents occur when a vesselis maneuvering, such paragraph 70; Caccia's Proposed Findings at 1-2). Much of the testimony as at docking or anchoring facilities, rather than w hen the vesselis traveling of Intervenors' witness Dr. Fisher was directed toward this contention." To i

at a steady speed (ibid.; Tr. 3475). In the 24-mile river section of interest, begin with, Dr. Fisher testified that an LNG vessel has more " sail arca" e

loaded LNG tankers will be proceeding upstream and will not be engaging than,ther vessels of comparable draft because its freeboard is as much as in maneuvers. In addition, they will be under Coast Guard escort and ac-four times larger than that of a comparable tanker (Fisher Prepared companied by one or more tugs, which would assist should the ship become Testimony at 12-13). Consequently, it is more susceptible to being thrown disabled. in Wew of the evidence, we find that loss of propulsion by an LNG off course by lateral winds; thus, in the Delaware River it might be subject i

tanker.in the river section of interest n unlikely, and should i, occur the to groundings with greater frequency than other vessels (ibid.)." Com.

resultant risk would be reduced by the assisting escort vessels. fr. any case, mander llenn testified, however, that he did not see the tv ge freeboards of the Applicants' analysis includes accidents resulting from ioss of propulsion LNG ships as being unique because there are other !na,e vessels that have (Kaleikar Supplemental Testimony at 16). Thus, there is no merit to Dr.

similar large sail areas, either by design or by the way they carry their cargo Fisher's contention that the methodology to determine risks does not en-r (Tr. 3470). Lieutenant Stanton testified that he knew of only one case in compass such possibilities (Fisher Prepared Testimony at 14).

which a ship with a large sail area was blown off course in the Delaware: an

47. Dr. Fisher further testified that the greater horsepower of an LNG automobile carrier, with shallow draft, was blown about the channel by vessel, which is about twice that of tankers having comparable draft, does gusty winds in a thunderstorm and requited assistance from a tug (Tr.

not add to Ile ship's maneuverability, except in the limited situation where 3473-75). lie characterized is as "a remarkable incident," and went on to the pilot v.shes to accelerate the vessel forwani to avoid an accident (Fisher say that the Coast Guard would not permit an LNG vessel to travel on the Prepared l'estimony at 15). Captain Van Leuven testified, however, that at river if a severe weather warning was out (Tr. 3474). The lloard notes, first, low forward velocities, high horsepower applied in short bursts with the that vesseis with large sail areas hat e been. plying the Delaware with no ap_

rudder hard over would change the direction of the ship without increasing parent higher incidence of risk thati other large ships. Further, LNG vessels its speed to any great extent (Tr. 3714). lie also testified that the bow will be equipped with bow thrusters, w hich, according to Commander ifenn thrusters cold be used to alter the ship's direction at speeds from 0 to 8 and Captain Van Leuven, Master of the S.S. LNG AQUARIUS, would aid knots (Tr. 3713)." Dr. Fisher also testified that the steam-turbine driven the vessel in maintaining stability under conditions of strong lateral winds LNG ves;els have very slow response characteristics (Fisher Prepared (Tr. 3483, 3722). Finally, LNG vessels would be accompanied by at least Testimony at 16). On cross-examination, howeser, Dr. f isher said that the one tug, which could assist in the event of wind displacement (Tr. 3472). In view of this evidence, we reject the contention by the Intervenors that the

..p,, ri,. cr offered extensive testimony on the cIIcc'iveness of bow thrusters, all of which s

higb freeboard of LNG vessels would make them more susceptible to we have considered but m 4 all of which need be reviewed here. IIis pousion was summarized when he said,"[Djetween 0 and 3 knots they are very effective. Hetween 3 and 6 knots their ef.

fcctiveness is appreciably reduced. Above 6 knots they are essennally useless" tTr. 3566). Ilis "Ihe Board notes ihat Dr. Fhher lessihed that he had donc no numerical analysis and had teslimony was contradicted by that of Captain Van Leuven, however, as just ind6cated. Cap-seached "quatuurwe conth uns" (Ir. 3638; I hher PreparcJ Testimony at it, emphash in tain Van Leuven testified further that at 8 knots "there is enough ef fectiveness in the bow

,,ig;g3, thruster on the LNG AQUARitlS to significantly move the bow in one direcuon or the other"

" T" ' post his testimony with regard to wind-caused accidents involving large ships, Dr.

(Tr. 3783). On considering Captain Van Leuven's experience compared to t)r. Fisher's, we give I hhes. ; some acudent casn h>ted in the l odd Shipyard report, " Nature of Ship Colhsions more weight to the testimony of Captain Van t euven isce Tr. 3708 W. 3418-23). Th-In-Within Posts" th'hher Prepared Testimony at 13; Applicants

• Es.12). Dr. thher sup-tervenors argue that the bow thrusters of the LNG AQUARIUS have capabilities not found plcmented his prepared test many catensively under crowenamination and questioning by the generally in bow thrusters on LNG vessels (Intervenors' Propovd Findmgs, paragraph 67). We Board ( Tr. 3414-37). We have studied all of ihis eviderke carefully and believe that is speak s han found no evideswe to suppora this contention. Indeed, the LNG AQUARIUS is one of a for itself; herwe we have decided that we need not review it here. We Lnd that the evidence does class of LNG tankers being bush by General Dynamics, a Deet af which woutJ serve the West nos support the interpretation uhith Dr. Fisher placed on the accident reports he cited from the Deptf rd LNG,ermmat should it 30 into operation (Tr. 3715; Board Ex. 2 at 169-170).

Todd repon.

661 660

contention that misjudgments, misunderstanding, or misapplication of res'ponse time of a large steam-turbine drive system would be about 60 regulations will negate the safety function of the rules in question. With seconds, compared to about 40 seconds for a large diesel direct-drive regard to Dr. Fisher's claim that the procedures promulgated in the COTP system. The bow thrusters of the LNG AQUARIUS have a response time of order for LNG / LPG vessels have been followed by pilots for the past 10 about 10 seconds (Tr. 3720-21). While transiting the Delaware, LNG vessels years, we note that the order requires several procedures which are not will normally travel at 12 to 15 knots except when meeting another vessel; in followed at present by conventional large vesni traffic. Conventional large meeting situations their speed will be restricted to less than 12 knots, to give vessels do not lave Coast Guard and tug escorts. They do not have speed a relative meeting speed of 12 knots (Kaleikar Supplemental Testimony, li,aits in meeting situations. Other river traffic is not forbidden to overtake Appendix B; Tr. 3439-40,3445)." These facts suggest that at speeds above 8 them. They do not have to notify the Coast Guard and obtain prior ap.

knots, the greater horsepower of LNG tankers makes them more proval before transiting the Delaware (Katelkar Supplemental Testimony, maneuverable than conventional large vessels, at least in some situations.

Appendix B; Fisher Prepared Testimony at 19-20; Tr. 3541-43). We believe At speedi less than 8 knots, the bow thrusters, and under certain cir-that these procedures, which are required in the COTP order for LNG / LPG cumstances the greater horsepower, makes lag ships more maneuverable vessels and are over and above those in current use by conventional vessels, than conventional vessels. Applicants' witnest Captain Knapp, a Delaware will be of major importance in reducing the risk of accident. We reject In-I River pilot, testified that greater horsepower, collision-avoidance systems, tervenors' contention that the Applicants' data base has already accounted and bow thrusters would provide an added margin of safety for LNG for such procedures. We conclude, on the basis of the foregoing, that the tankers over vessels which did not have such equipment (Tr. 3574). We proedures contained in the COTP order for LNG / LPG vessels on the believe that the evidence indicates that the operating characteristics of LNG Delaware River will reduce the accident risk for such vessels, relative to the vessels do provide them with collision-avoidance capabilities not possessed accident risk for conventional large vessels. This fact imparts conservatism by conventional ships. We find that the performance capability of LNG to the accident rate estimate calculated by the Applicant.

tankers, which was not factored into the collision probabili!y calculation of

49. In conclusion, we have found that the accident rate calculated by the the Applicants, imparts conservatism to the estimate.

Applicants,1.51 x 10+ collisions per mile-year, is reasonable. We also have

48. Dr. Fisher also testified that it is erroneous to rely on the Captain of found certain conservatisms in the method used by Applicants to screen ac-the Port of Philadelphia order, " Procedures for the Movement of cidents in the casually data bav. We further find, in view of the foregoing LNG / LPG," as a conservatism, because pilot misjudgments, qualitative arguments, that the realistic probability has been shown to be misumferstandings, or misapplication of regulations will tend to negate lower than the calculated probability and, consequently, the Applicants' their intended safety functions (3 isher Prepared Testimony at 17-19; Tr.

calculation of accident tate is consersative pursuant to guidelines contained 3541-45). lie also contended that the procedures promulgated by the COTP in the NRC Standard Review Plan (NUREG-75/087,52.2.3 (1975)).

order for the operation of LNG tankers on the Delaware River are not new; he said that the Delaware River Pilots Association has becu following essen-tially the same rules over the past 10 years (Fisher Prepared Testimony at 19-20). Lieutenant Stanton testified that LNG and LPG vessels would tansit C. Spills Per Collision the Delaware River under the surveillance of a Coast Guard escort vessel (Tr. 3482). A pilot or master who violates the order of the Captain of the

50. The Appeal Board questioned the assumptions made by Applictnts, Port would be subject to a 580,000 civil penalty, or a criminal penalty if the and accepted by Staff, regarding the angle of collision and relative speed of violation were willful (ibid.), in his opinion, the presence of the Coast colliding ships and called for the Applicants to provide additior:al founda-Guard encourages compliance with the rules (Tr. 3483). Captain Van Leuven tions for these assumptions using actual ship experience (ALAB-429,6 testified that he believes that the promulgation and use of the COTP rules NRC at 239-240). In addition, the validity of the Minorsky calculation, for LNG / LPG vessels would improve the rafety of navigation for such which was used by the Applicants to predict the depth of penetration into vessels (Tr. 3712). The evidence indicates that the Coast Guard will insure the LNG tanker by a colliding ship, was questioned by the Appeal Board that LNG vessels adhere to the regulations. We do not accept Dr. Fisher's (id, at 240). We turn now to a consideration of these questints and to the determination of the probability that an LNG spill would result from a col-

• • T he Coast Guard procedures for tNG/ LPG vessels state that normally in a raceting situa.

Itsion of an LNG tanker with another ship in the river segment of interest.

toon. both vessets should proceed at 6 knots (Kaleikar Supplemental Testimony, Appendia B).

663 662

51. According to the Applicants, any collision which might occur in the but these types of accidents are not relevant to the situation we are consider-catchment zone would occur at narrow angles, seldom exceeding 10' to 15 *,

ing (Kalelkar Supplemental Testimony at 34; Read Supplemental Testimony because the LNG tanker would be moving in a channel which (over most of at 23-24). The assumption that impact speeds are uniformly distributed bet-the distance) is only 800 feet wide and has no docks or moorings, and the ween 0 and the maximum speed at which ships transit a given region svas tanker would be accompanied by Coast Guard and tug escorts (Kaleikar also made in the collision analyses by Bovet and by SAI (Kaleikar Eup-Supplemental Testimony at 35).*' At bends, notably Bulkhead Bar where plemental Testimony at 34).** While data are available to support this the channel widens to 1,600 feet, a collision angle of 40* is considered possi-assumption, they are inadequate to provide statistical validation (Tr. 3036).

ble (ib;d.).85 However, a collision at a bend is thought to be very unlikely

53. Staff chose not to use a posteriori methods, such as the hiinorsky because the COTP rules for LNG tankers do not permit meeting another analysis, to derive a probability of spill estimate, because relevant data are ship at a bend (ibid.; Tr. 3037-39). In applying the hiinorsky analysis, Ap-insuft'icient (Read Supplemental Testimony at 23-24). Rather, asing an a plicants assumed that collisions were uniformly distributed between 0* and priori technique, Staff concluded that the Applicants' assumption of 45' (ibid.). With regard to actual ship experience, only data from the uniform independent probabilities for impact angle and speed are conser-24-mile section of the Delaware in queuion would be relevant to the instant vative because:

analysis, and no angle-of-collision data are available for the few collisions (I) Observed channel collisions are almost invariably at much smaller which have occurred in this section of the river (id. at 35-36).

relative velocities than the maximum 12 knots because of corrective

52. With regard to speed, Applicants assumed that the relative selocities action undertaken by the respective vessels.

of an LNG tanker and a colliding ship would be uniformly distributed be-(2) Large angle turns by large ships within the river channel are not ex-tween 0 and 12 knots (id. at 34). Twelve knots ws.s selected because that is pected within the 24-mile area of concern because of the configura-the maximum relative speed permitted by COTP regulations for an LNG tion of the river channel.

tanker and another ship in a meeting situation (Tr. 3036). Lower speeds (3) All other accidents besides collisions are inherently unlikely to were included in the distribution because ship masters tend to do everything cause deep penetrations.

they can to minimize the severity of impact, once they become aware that a collision is impending (Kaletkar Supplemental Testimony at 34). Staff (Id. at 24-28.)

witness Dr. Read testified that "lo]bserved channel collisions are almost in-variably at much smaller relative velocities than the maximum 12 knots

54. Intervenor'-witness Dr. Fisher criticized the use of hiinorsky's because of corrective action undertaken by the respective vessels" (Read model, on the grounds that the model was not applicable to LNG tankers Supplemental Testimony at 28). Little historical data exists on actual vessel "because the collision characteristics of LNG carriers are different than velocities in low. speed collisions (id. at 23; Kalelkar Supplemental (sic) those of the vessels studied by hiinorsky" (Fisher Prepared Testimony Testimony at 34; Tr. 3035-36). Such data are available for very severe colli-at 72). Dr. Fisher also testified that he had been told by hir. hiinorsky, in a sions because severe collisions tend to be carefully scrutinized afterwards, telephone conversation, that the hiinorsky calculation cannot be applied to double-hulled LNG vessels or to collision angles less than about 60* or 70*

in recent testimony the Apphcants have used the notation 0* for head on collisions and 90 (Tr. 3630). Upon cross-examination, however, Dr; Fisher acknowledged for beam-on cothsior.:. Earlier. Applicants used htmorsky's designation of beam-on colhsions as 0 and head on colhsions as 9a* (Applicante tm.10 p. 3). The Appeal Board adopied the that hiinorsky had himself applied his correlation to LNG ankers (Tr.

lauer convention in ALAB429 (6 NRC at 239. n. 58). Because most of the testimony which we are 3635-36)." In rebuttal testimony, Applicants' wimess Dr. Kalelkar said that reviewing uses the former notation thead on 0*; beam-on. 90*) we are adopting that conven-he and members of his staff had visited and consulted with hir. hiinorsky, tion in this decision.

and had shown to and discussed with hiinorsky the Apph. cants' colh..ston "htr. Caccia claims that the ship channel makes a 70* bend at Bulkhead Bar Range (Caccia Proposed Findings at 1). hvoreover. under cross-cuamination Dr. Kalenar appeared io iesiify analysis (Tr. 3697-98). According to Dr. Kalelkar, hiinorsky said that his that New Castle Range and Deepwater Point R2nge meet at an angle of 70*(Tr. 3075). In fact.

methodology would predict adequate results if applied at narrow angles and Bulkmad liar Range is interposed herwern New Castle and Deepwater Pome Ranges, so that a ship n.oving between these two ranges must turn to enter Bulkhead Bar Range and then turn "D.ht. Bovet. "Prehminary Analysis of Tanker Grounding and Collisions." U.S. Coast again to leave it. Neither of these turns exceeds 40'(see photocopy of section of NOS chart Guard (January 1973); see n. Ig, supra, for reference to SAI study.

1238 8 attached to letter from Captain Wiman to Chairman 1 uton, dated November 10,1977).

"This study by htmorsky was carried out for klarathon O I Company and involved Cooks TLne is no 70* bend in the sh.p channel in the 24-mile river section of interest.

Inlet. Alaska, and Negishi, Jaaen (Tr. 3 706).

664 665

that'the numbers being obtained by the Applicants were in the same range the inboard distance of the LNG tank wall (approximately 4 metecs) that hiinorsky, himself, had obtained when he applied his methodology to (ALAB-429,6 NRC at 240). Dr. Katelkar testified that Bovet's study was LNG tankers (ibid.).

biased toward major collisions, in which detailed casualty investigations

55. The Board received into evidence a copy of a paper published by had been made, and thus wu biased toward relatively great penertration hiinorsky in 1959, in which an attempt was made to predict what structural depth (Kalelkar Supplemental Testimony at 37). Consequently, the average strength should be built into the hull of a nuclear-powered ship in order to depth of penetration in the cases studied by Bovet cannot be used as a safely absorb a given amount of kinetic energy in a collision (Applicants' criterion for the spill probability analysis carried out by the Applicants Ex.13, p. I)." The analytical method developed in the paper also permits (ibid.). h1orcover, it was shown that Bovet's method of prediction was less computtiion of the maximum depth of penetration of a striking ship, given accurate, or at least less precise, than hiinctsky's method (id. at 39-41). Dr.

known characteristics and structural features of the colliding ships (id., p.

Kaleikar also attempted, but without success, to determine the source of the 4). Our 5tudy of this paper has shown that hiinorsky eliminated sharp angle co:iision data used by Comstock nnd Robertson (id. at 3d). Because their collisiuts from his analytical analysis because (1) use of them would have study focused on ship structtee and design which govern survivability of rendered the selection of significant strength structural members too dif.

ships struck in collisions, D:, Kalelkar believes that Comstock and Robert-ficuit,(2) eliminating them minimized " error due to neglecting components son also selected a data base of major collisions rather than a spectrum of of kinetic energy parallel to the struck ship's axis," and (3) penetrations in collision events representative of the entire range of possible collisions.

oclique collisions were much smaller than in right a.sgle collisions and Thus, tbc average penetration depth in the study by Comstock and Robert-therc6pc oblige collisions were not pertinent to the problem of protecting son is likewise not applicable to the instant case (ibid.). Dr. Read also the suuck ship's nuclear plant (id. at 1). We find nothing in the paper to testified that the studies by Bovet and by Comstock end Robertson are not cause us to believe that hiinorsky's method is invalid when applied to obli-directly comparable to the instant spill probability problem for LNG que collisions. The paper leads us to believe that the accuracy of'the hiinor.

tankers (Read Supplemental Testimony at 25). He does not believe that sky prediction declines as one moves from a 90' collision toward a 0* colli-there is a disagreement between hiinorsky's work and that of Hovet and that sion. Futhermore, the paper makes it quite clear that in a right angle colli-of Comstock and Robertson (id. at 26).

sion, the hiinorsky correlation is very good for high-energy collisions but

57. The Appeal Board indicated that a determination should be made as much less so for low-energy collisions (id. at 3-4). hiinorsky defined a high-to whether there is a difference in the spill vulnerability between an LNG energy collision as being of the order of 10 to 16 knots (id. at 3). With ship with freestanding tanks as opposed to one with membrane tanks reference to the Applicants' analysis, obsiously the higher speeds of the0 to (ALAB-429,6 N RC at 246). Freestanding tanks are self-supported and have 12-knot distribution are in the range considered to be high-energy collisions sufficient strength in themselves to withstand cargo loads and the stresses of by hiinorsky. 'ligh-energy collisions are, of course, more likely to cause a cargo movement (Kalelkar Supplemental Testimony at 44-46; Tr. 3454-55, cargo spill from an LNG tanker than lonnergy collisions." Thus, for our 3456-69). hiembrane tanks have a thin, metal barrier which contains the purposes, it is important that the hiinersky model predicts accurately at LNG and which is supported by an insulation system which, in turn, is sup-high speeds and not so important that it predicts less accurately at low ported by the inner hull and sidewalls (ibid.). The freestanding systems in-speeds. Finally, we note that we have no information about how inaccurate clude prismatic, cylindrical, and spherical tank-designs, while the mem-the model *,:comes as the angle of collision decreases, i.e., moves away brane systems usually use a prismatic waffled configuration (Kaleikar Sup-from 90* toward 0*

plemental Testimony at 46-47). According to Commander llenn, the Coast

56. The Appeal Board questioned the validity of the Minorsky calcula-Guard considers one system as safe as another (Tr. 3468-69). The Ap-tion because collision studies by Bovet and by Comstock and Robertson plicants and Staff, however, believe that ships with prismatic tanks would, gave average penetration depths of approximately 5 meters, which exceeds under most situations of marine traffic, be more susceptible to a spill than ships with cylindrical or spherical tanks, because prismatic tanks have larger

V.U. Minorsky. "An Analysis of Ship Colhsions With Wference to Protaction of Nuclear areas close to the hull than cylindrical or spherical tanks (Katelkar Sup-Power Plants." Journal of Ship Researth (October 1959).

plementalTestimony et 46-47; Read SupplementalTestimony, Appendix A, "According to the FPC analysis. a vessci of 38 mu tons displacement would have to impac:

an LNG lanker at a speed in excess of b8 knots. assuming an impact angle of 90*. in order to

p. 6; Tr. 3277). On balance, we believe that the evidence supports the view rupture a cargo tank (Board Ex. 2, p.191).

of Applicants and Staff. In any case, however, the Applicants have factored 666 667

t both tank designs into their hiinorsky analysis, and therefore the spill pro-clude, with regard to the coatradictory evidence concerning hir. Minorsky's bability is valid for both designs (Applicants' Ex.10, pp.1-2).

comment about Appli< ants' use of the hiinorsky model, that the greater

58. In applying the Minorsky analysis to the LNG problem, the Ap.

weight must be given to the testimony of Dr. Kalelkar. We believe that the plicants assumed that if the striking ship penetrated to a depth equivalent to Minorsky model, while not ideally applicable 40 the instant problem, is the inboard distance of the LNG tank, the tank would be ruptured and a nevertheless a valid, reasonable, and adequate method for application here.

rapid loss of cargo would result (Applicants' Ex. II, p. 24). An additional We believe it is to be pieferred over Bovet's method. Indeed, it may be the assumption made in using the Minorsky model is that all of the (nergy of most appropriate technique available for the Applic at's analysis (r.ce impact is absorbed by the struck ship (Tr. 2681,3698). Corsidered together, Board Ex. I, p. 217).

we believe inat these two assumptions lend conservatism to the probability

62. With regard to the assumed distributions for collision angle and calculation.

speed at impact, we find both assumptions to be reasonable. The evidence

59. Using the Minorsky calcu!ation, Applicants obtained a probability shows that LNG ship size, physical characteristics of the channel, COTP of 5 x 10-8 spills per collision for 23 miles of the 24-mile river section of in-rules for LNG tankers, Coast Guard superv.ision, and tug escort will make terest, and 5 x 10 2 for the I-mile section near the C & D Canal, where 90*

the likelihood that an LNG ship would be impacted at angles greater than collisions are possible (Applicants' Ex. II, p. 24). The value 5 x 10 8 is con-45" negligible, except in the vicinity of the C & D Canal. Applicants have sistent with the estimate of 6 x 10 ' spills per collision obtained by the FPC reasonably accounted for the likelihood that wider angled collisions could for collisions at piers, harbors, and entrances (Board Ex. 2, p.198). The occur there by assuming that collision angles in that section of the Hvei FPC estimate is expected to be greater than Applicants' estimate of 5 x 10 8 are distributed between 0* and 90*. We find it reasonable to assume that because it applies to the entire river and hence considers all impact angles impact speeds are uniformly distributed between 0 and 12 knots because (Board Ex. 2, p.180-83). The fact that it is not as great as Applicants' of COTP rules, which require that the relative meeting speed of an LNG estimate of 5 x 10 2, applicable to the vicinity of the C & D Canal where all tanker and another ship not exceed 12 knots, and the presence of a Coast impact angles are possible, can be attributed to differences in data base and Guard escort to enforce those rules. Considering the Coast Guard escort, analytical method. The larger value obtained by the Applicants for allim-the presence of a lug, the collision avoidance capability of LNG ships, and the pact angles is consistera with the belief that the Minorsky method is conser-evasive maneuvers wh;ch pilots are known to take when they become vative.

aware that a collision is imminent, we conclude that impact speeds would

60. We have noted tha: Staff chose to avoid a posteriori methods in be more densely t'istributed at lower speeds in the O to 12-knot range.

eva'uating the Applicants' estimate of spill probability, because of the Thus, we think that this assumption is conservative. We also firid the paucity of historical data. Nevertheless, in the end Staff accepted the prob-Minorsky model to be conservative because of the assumption that all of ability value obtained by Applicants from the Minorsky analysis, "not the energy of impact is absorbed by the struck ship. We conclude that because it was likely to be correct, but because there was no basis to believe Applicants' estimate of spill probability for 23 miles of the catchment that any accident that was predictable near ArtificialIsland would cause the zone,5 x 10 5 spills per collision, is both reasonable and conservative. The rapid release of LNG gas necessary to endanger the nuclear power estimate for the I. mile section in the vicinity of the canal,5 x 10 2, is like-plants..." (Read Supplemental Testimony at 28-29).

wise reasonable and conservative for the same-reasons. We accept these

61. Joint Intervenors argue that the Applicant's estimatt of spill prob-estimates.

ability is "neither conservative nor even necessarily axcurate.. "

(Intervenors' Proposed Findings at 38). They cite the testimony of Dr.

D.

Vapor Clouds Per Spill Fisher, wherein he developed scenarios for collision angles greater than 45 *

(id. at 40; Fisher Prepared Testimony at 12-22, 29-33). We have reviewed

63. The Appeal Board found insufficient support in the record for the Dr. Fisher's testimony in paragraphs 11 and I'!,20 through 22,27, and 45 estimate by Applicants and Staff that the probability of a flammable through 48, and have rejected so much of it as pertains to the cloud of LNG not igniting, given a spill, before it reached the llope Creek maneuverability and control of LNG tankers in the river section of interest.

site was 0.1 (ALAB-429, 6 NRC at 241). Since there has never been a We have also reviewed Dr. Fisher's testimony on the Minorsky model in large accidental release of LNG in a maritime casualty, the probability of paragraph 54. Out own study of Minorsky's 1959 paper causes us to con-a spill not igniting immediately cannot be evaluated from historical evi-669 668

de'nce (Kalelkar Supplemental Testimony at 49). There have been how-land SAI obtaic.ed an ignition probability of 0.94 by the time the vapor ever, more inan 100 reported vapor cloud relea se accidents on land, in-clouds had dispersed over a surface area of about 105 square meters volving a diversity of hydrocarbons and release conditions (Read Supple-(Kalelkar Supplemental Testimony at 50-51; Staff Exhibit 1-F, p.10).

mental Testimony at 29). In over half of these cases ignition was immedi-f LNG tankers of the type to be used by Tenneco, having a capacity of ate, and in only a very few did the gas disperse prior to ignition. These 125,000 cubic meters, have a net surface area in excess of 105 square observations suggest that the probability of ignition immediately upon meters which, in a severe collision, would contcin a high density of igni-release is more than 0.9 (ibid.). To apply this probability to the maritime tion sources (Kaleikar Supplemental Testimony at 51). Finally, the FPC, problem being considered here, however, requires the demonstration that in its DEIS for the proposed West Deptford LNG terininal, estimated conditions for ignition in a marine accident in the Delaware are at least as ignition probability conservatively to be 0.90, although the Appeal Board good as in a lan 1 based accident (ibid).

has complained that the FPC estimates are based on "the qualitative or

64. The energy that would be dissipated by a large ship strik, an intuitive judgment of experienced persons" (Board Ex. 2, p.194; ALAB-LNG tanker with sufficient force to penetrate the double hull and rupture 429,6 NRC at 241, n. 72).

a cargo tank is tremendcus; it has been estimated to amount to over 10'

66. Intervenors' witness Dr. Fisher criticized the Applicants' estimates Joules (Read Supplemental Testimony at 30). This dissipation of energy of ignition probability on the ground that "the foundation of the ten per-would give rise io substantial heating of metal surfaces and frictional cc..t probability of vapor cloud formation is, in this area, essentially ' sci-sparking during impact (Kalekar Supplemental Testimony at 49). Follow-uce by concensus' (sic)," (Fisher Prepared Testimony at 33). Dr. Fisher ing impact, electrical sparking would be expected from severed and also criticized Applicants and Staff for not producing new evidence to shorted electrical cables, and frictional sparking would continue to be support the ignition probability value, on the grounds that new evidence produced as the ships rub together (ibid.). The use of spark igniters in gas was called for by ALAB-429 (id. at 245). In fact, the Appeal Board found appliances attests to the efficacy of spark-ignition sources for flammable that there was " insufficient support in the record for the assumed prob-gas (Read Supplemental Testimony at 30). Both Applicants and Staff ability value" and asked "that a greater effort [be] made to arrive at c believe that such sparking virtually assures that the methane cascading reasonable estimate of that probability" (ALAB-429, 6 NRC at 241).

from a ruptured cargo tank of an LNG tanker and mixing with air would Clearly, to the extent that new information is avaifable, Applicants and be ignited (ibid.; Kaleikar Supplemental Testimony at 49)." Thus, Appli-Staff should be expected to make use of it in response to the Appeal cants believe that a 0.99 probability of ignition is realistic, but conserva-Boaid's request, and to the extent that rew information is not avaibble, tively estimates the probability at 0.9 (Kalekar Supplemental Testimony at we would vt them to apply " greater effort" in using what has already 50).

been pr aduced. We believe that Applicants and Staff have adequately

65. Other evaluations of the probability of ignition of LNG released complied with the Appeal Board's request. Further, we see no merit in f om a tanker involved in an accident support the conclusion of Appli-Dr. Fisher's allegation that the agreement in ignition probabilities among cants and Staff. In the SAI risk analysis for the proposed Raccoon Island different analyses represents " science by consensus." If the probGility of LNG terminal, an investigation was made of 12 tank ship and tank barge ignition is really 0.90 or greater, thu we would expect estimates from dif-accidents in which a low flash point product was released (Arvedlund j

ferent studies to be 0.90 or greater. The Intervenors have produced no Prepared Testimony at 6; see Board Ex.1). Ignition occurred immediately evidence to support their contention that the 0.90 ignition probability is in 11 of these cases. In the one accident in which ignition was not immedi-not conservative, nor does the record indicate that this probability value is ate, the cloud was ignited a few minutes later (ibid.). From these data one i

arbitrary (see Fisher Prepared Testimony at 34). We believe that the evi-can estimate the probability of immediate ignition to be 0.92. In another dence indicates the ignition probability actually is greater than 0.90.

study involving propane released in truck, pipeline, and rail accidents on Therefore we conclude that the Applicants' estimate for nonignition prob-ability,0.I, is both reasonable and conservative.

"There has been one marine accident in U. S. waters in which an unignited gas cloud was l

67. The Appeat Board requested, further, that consideration be given released (Read Supplemental Testimony at 30, n. 5). In that incident a barge carrying the likelihood and consequences of a fire involving the entire cargo of an nrrigeraica propylene sank witinout its cargo tank being penetrated, and the spill occurred LNG ship in the river near the plant (ALAB-429, 6 NRC at 246). The under water. The circumstances associated with this accident are unlike those postulated for a spilt from an LNG tanker.

670 671

sp'ill would result in a fire which would spread to the rest of the cargo Board (LBP-77 22, 5 NRC 694; ALAB-429,6 NRC at 242). Defined as either (1) instantaneously, immediately producing a pool fire from five the piobability that an LNG cloud formed in any 1-mile section of the tanks, or (2) in domino fashion, such that the second tank is ignited as catchment zone would be transported by the prevailing wind to the llope fuel from the first is consumed, the third is ignited as fuel from the Creek site and would arrive at the site in a flammable concentration, the second is consumed, and so on, withjhc result that the fire lasts five meteorological factor was calculated from data collected on Artificial times as long as a one-tank fire (Kaleikar Supplemental Testimony at 51-Island assuming a spectrum of meteorological stability conditions (Appli-52). Fire size as a function of time, radiation emitted by the fire, and cants' Ex. II, p. 23-28). No additional evidence has been introduced to hazard distances associated with the fire were calculated using a method cause us to change our finding that the meteorological factor accepted in reported in a paper by Ray and Kalelkar (see Kaleikar Supplemental Testi-LPB-77-22 and ALAB-429 is reasonable. We accept 0.35 as the prob-mony, Appendix E)." The results showed that concrete buildings would ability that an LNG cloud would be transported to the llope Creek site be safe at distances beyond 900 m from the center of the fire (Kalelkar from all 1-mile segments of the river except where the C & D Canal joins Supplemental Testimony at 52). The 11 ope Creek plant would be about the Delaware River, and 0.0M for the probability of a cloud being trans-2,000 m from the fire center and hence would not be affected by the fire ported from the vicinity of the C & D Canal to the plant site (see Kaleikar's (ibid). Staff witness Read likewise testified that a massive lire storm Supplemental Testimony at 53)."

caused by a multi-tank release of LNG would not constitute a significant

71. Intervenors' Witness Dr. Fisher expressed concern that progres-threat to the llope Creek plant, because the concrete walls of the plant sive fadure of tanks might occur as a result of damage to noncryogenic could withstand the radiant flux from such a fire (Read Supplemental structures of the ship by LNG spilled from one tank; he suggested that Testimony at 31).

such a sequence of events might enlarge the catchment zone (Fisher Pre-

68. The Joint Intervenors, in arguing about a multi-tank disaster, do pared Testimony at 22). Dr. Read testified, however, that it is the rate at not address themselves to the five-tank fire scenario about which the Ap-which LNG is spilled, not the number of tanks spilled, that is of signifi-peal Board expressed concern (Intervenors' Proposed Findings at 58).

cance in determining the catchment distance (Tr. 3226). In formulating Rather, they postulate a multi-tank spill followed by detonation of the the one-tank spill scenario, it has been assumed that the spill from the methane cloud (ibid). To support their argument, they cite testimony of ruptured tank instantaneously releases the entire contents (ibid.). This as-Dr. Read given in cross-examination, but they have introduced no evi-sumption gives the 24-mile catchment zone and, in Dr. Read's view, is it-dence of their own on this subject (ibid.). A full reading of the transcript self conservative (Tr. 3229). Should the contents of all five tanks reveals tha: Dr. Read testified that detonation of spilled LNG would be somehow be released very rapidly into the hull, the boiling of the LNG very unlikely, and in the event it did occur, it would cause a weak blast would destroy the ship by overpressurization, and the probability of igni-which would not propogate sufficient energy over the distance to the tion would be very high (Tr. 3230-31). Dr. Kalelkar testified that fne plant to cause overpressure that would damage the plant (Tr. 3286-91).

structure of the hull and the presence of water outside to act as a heat This question has been considered previously, however, and disposed of source for the spilled LNG would make a multi-tank spill caused by cold by us; it need not be reconsidered here (LPB-74-79,8 AEC at 751).

stress, as suggested by Dr. Fisher, an incredible event (Tr. 3085-87).

69. Upon the evidence, we conclude that a multi tank fire on an LNG Commander llenn testified that the Coast Guard considered a one-tank ship in the river would not threaten the llope Creek plant. We find that failure as the only credible accident for an LNG tanker (Tr. 3481). The such an accident need not be factored into the risk analysis.

Coast Guard position on this question is based on "the judgment of many Coast Guard officers with varied backgrounds, both in the operation of E.

Meteorological Factor vessels, in the design of ships, and in the investigation of casulatics" (Tr.

70. The meteorological factor, previously accepted by this Board, was "Previously the Applicants used 0.002 for the meteorological factor for the portion of the found to be reasonable and approrpiately conservative by the Appeal river in the vicinity of the C & D Canal because this value was obtained for mile N7 where the canal joins the Delaware River (Applicants' Ex. II, p. 29). It is not clear why Appli-

"P. K. Raj and A. S. Kaleikar Fire llazard Presented by a Spreading Burning Pool of cants now use 0.004. Since 0.004 is more conservative than the previous estimate, however, LNG on water," Paper No. 73 25, Western States Section of the Combustion Institute, and both we and the Appeal Board have found 0.002 acceptable, we have no difliculty in October 1973.

now ea.cepting 0.004.

s 672 673

I TABLEI 3485-86). While this view has not been proven beyond all doubt, it is rup-ported by tests sponsored by the Coast Guard and other agencies (Tr.

3486). Upon consideration of the foregoing evidence, we conclude that a Applicants' and Staff's Estimates of Parameter Values and Resultant Prob-abilities for LNG Traffic Over 24-Mile Range multi-tank spill occurring at a rate which would produce a cloud large enough to increase the catchment distance, without the cloud igniting at Applicant Staff the ship, is so unlikely that it need not be considered in the risk analysis.

• F.

Probability of a Flammable LNG Vapor Cloud Reaching the Tanker Trips Plant Site f

292 360 per Year

72. The probability that a flammable gas cloud released by an accident Collisions involving an LNG tanker in the Delaware River will reach the site of the 1.51 x 10 1.5 x 104 Ilope Creek plant is given by the product of the conditional probabilities per Mile-Year discussed above." Based on its revised and updated analysis, the Appli-cants have calculated the overall probability of an LNG vapor cloud pre-Probability 0.005 (24-mi) 0.005 senting a potential threat to the hope Creek plant to be 8.6 F 10 ' occur-of Spill 0.05 (C & D) rences per year (Kalelkar Supplemental Testimony at $3). Initially, the Staff chose to rely upon its original analysis, which gave an overall prob.

Probability of 0.1-0.1 ability of I x 104 occurrences per year (rounded up from 9.5 x 10 8)(Read Nonignition Supplemental Test %ony at 37-38; Staff Exhibit 1-F, p.14). These prob-ability estimates and the values from which they were calculated are Meteorological 0.35 (24-mi) 0.35 Factor 0.041 (C & D) summarized in Table I (p. 675). The Staff subsequently has modified its position, however, and now adopts the Applicants' estimate (Staff's Pro-posed Findings, paragraphs 98-99).

Product 8.6 x 10 '

9 5 x 10 8

73. We have found all of the Applicants' estimates for the factors used to calculate the overall probability to be reasonable, and therefore we fit.d the estimate of 8. 6 x 10' occurrences per year to be reasonable. In addi-tion, for reasons already given, we found Applicants' estimates for colli-sions per mile, spills per collision, and vapor clouds per spill to be conservative. On the basis of these findings, we could accept Applicants' estimate for the overall probability as a conservative estimate. We indi-mile, spills per collision, and vapor clouds per spill, along with Appli-cants' reasonable estimate for the meteorological factor. These values t.re cated, however, that Applicants' estimate of 292 tankers per year was rease table, while Staff's estimate of 360 tankers per year waa both rea-summarized in Table II (p. 676). The overall probability that can be sonable and conservative. We believe that a most conservative estimate calculated from them is I.I x 10' occurrences per year. We accept this can be obtained by using Staff's conservative estimate of number of value as a most conservative estimate of the probability that a flammable tankers per year and Applicants' conservative estimates of collisions per gas cloud, released in an accident involving an LNG tanker on the Dela-ware River (but not including a ramming of tower 97), will reach the "Dr. Read does not believe it.cl we and the Appeal Board should apply the term flope Creek plant. Because of the arguments that led us to find

"' "di'i "*I pr babihtics" to refer to the probabilistic factors for the sequential events conservatism in the estimates of tankers per year, collisions per mile considered a.n this (Read Suppicmental Testimony at 4-5). It appears that he would use spills per collision, and vapor clouds pcr spill, we believe that the realistic the term only in the specialized sense relating to conditionally distributed variates. We probability is lower than 1.1 x 10L Therefore we find this estimate to be beheve that applying the term te appropriate probabilistic factors in product rule calcula-ConservativC pursuant to NUREG-75/087,92.2.3 (1975).

lions is acceptabic usage. I.usever.

675 674

II. LPG TRAFFIC g

74. The Appeal Board found areas of uncertainty relating to LPG traffic similar to those relating to LNG traffic (ALAB-429, 6 NRC at n

3 g

l l

l l

x y

y 243). Specifically, that Board requested clarification as to the expected O

LPG traffic, an explanation of why the spill probability for LPG ships is not as large as that for LNG ships, and more information on how the i

3 flammability factor for LPG was determined (id. at 243-245). The Appeal H

o g

Board appears to consider LPG as " virtually" synonymous with propane, 2

N but indicated that it was also concerned about "other forms of river traf-o x

x

~

g j

q g

fic which could lead to flammable vapor clouds at the site, i.e., butane O

tankers" (id. at 243, n. 81). The Applicants have included in their analysis m

g a

propane, butane, butadiene, and vinyl chloride (Kalelkar Supplemental 3

Tesimony at $3-60). Staff, on the other hand, considered only " propane

,o o

and LPG" and " butane" (Staff Exhibit 1-F, p.14). The Coast Guard 2

2 E

o 8

5 defines LPG as " liquified petroleum gas including butane, butadiene, 3

j j

g y

propane, and propylene" (Kaleikar Supplemental Testimony, Appendix B, p.

~

C j

4 V-1). To comply with the Appeal Board's directive that we consider LPG, we hereby adopt the Coast Guard definition of LPG. We shall, therefore, y

g o

o i

consider the cumulative probability that a flammable cloud of propane, j

g, o

x gg o8 o

propylene, butane, butadiene, or LNG (methane) could reach the liope x

3 g

d6 66 C!

Creek plant as a result of an accident involving a tanker on the Delaware o

m h

g

~

8" River. Since the Applicants have also done a risk analysis for vinyl ch'.a-g ride shipments, we shall consider the evidence on vinyl chloride separately. V/c

. g

.g3 turn now to LPG.

E 4 4 4 4 y

eg cg A. Ships Per Year Sg Propane E

S 3

n

.E j

8 8g

75. In 1976 the Sun Oil facility at Marcus llook, Pennsylvania, was

~

x x

d y

J q

d6 66 m

~

completed, and this facility will begin receiving propane shipments ori a regular basis in the near future (Kalelkar Supplemental Testimony at 53;

~

E Tr. 3M2-43). When the project matures in 1980, the facility will receive a 3

maximum of 40 propane tankers per year (ibid.). Applicants use as their

.E g

3 y

estimate of propane tankers per year the number 40, based on Sun Oil's

7 E.

,0 x :E

• E y

projection of maximum number of shipments that could be received each x

t A

S6 5

5.h $

E year (Kaleikar Supplemental Testimony at 54). Staff, on the other hand, j,

t,0

.N 5 S j, SO o3 uses the estimate of 50 propane tankers per year, based on the maximum

$ x{

j [E.

j(

jm j{

jQ podble storage capacity of the Marcus flook facility (Read Supplemnval g

H U o.

0o Go eW Testimony at 9). It is normal practice, however, to maintain 20% to 25%

of storage capacity as a reserve, which accounts for the difference in the 676 677

estimates of Applicants and Staff. Staff considers its estimm of 50 plemental Testimony at 59). On this basis, Applicant estimates that ten tankers per year to be "a biased overestimate of future traffic rates" butadiene tankers will pass the liope Creek plant per year. Staff has made (Read Supplemental Testimony at 9). Considering the fact that 40 is the no specific estimate for butadiene traffic (Staff Exhibit I-F; Tr. 3219).

We conclude that the evidence indicates that ten butadiene shipments per maximum number of propane shipments that the Sun Oil facility could receive in a year, we believe that figure to be the more reasonable number l

year should be accounted for in the risk analysis.

for use in our analysis.

Propylene Total LPG Traffic

76. No evidence was presented to indicate that propylene is or will be

79. The total LPG traffic estimated by Applicants is $2 transits per shipped on the Delaware River. We conclude, therefore, that it need not yur, while Staff estimates 60 transits per year. The values which pro'.cc be considered.

these totals are summarized in Table ill (p. 680). We have found all of Applicants' estimoes reasonable except for the number of butane ship-Butsee j

reents. We have found Staff's estimate for butane to be reasonable, how-ever. If we use Staff's estimate for butane and Applicants' estimate for

77. The Applicants assumed that the number of butane carriers that the other LPG products, we obtain a total of 60 LPG transits per year wauld transit the Delaware each year would be two(Kalelkar Supplemental (see Table II, p. 676).

Testimony at 59-60). This assumption is based on the observation that

80. Joint intervenors argue that these estimates "are so uncertain as to there has been only one spot shipment of butane on the Delaware in the l

last 3 years, a shipment received by Gulf in 1977 (ibid.). The Staff, on the be nonconservative m themselves and to destroy the conservatism of the other hand, assumed that ten butane carriers would pass the llope Creek entire risk estimate" (Intervenors' Proposed Findings, paragraph 25c).

The basis for their argument is the increase in LPG traffic on the river plant each year (Read Supplemental Testimony at 9). The basis for Staff's judgment appeared to be current river traffic (ibid.; Staff Exhibit 1-F, p.

smce 1974, when Applicants estimated that there would be two LPG ship-5). But when Dr. Read was asked, in the evidentiary hearing, to explain ments per year on the river (id., paragraphs 23-25). Intervenors argue that the basis for Staff's estimate of ten butane tankers per year, he stated that Applicants and Staff have failed to account for the potential continued butane is used as a gasoline additive and is periodically shipped to re-increase in annual LPG shipments (id., paragraph 25d). The testimony of fineries on the upper reaches of the Delaware River (Tr. 3216)." The both Dr. Kaleikar and Dr. Read indicates, however, that the estimates of PP cants and Staff are based on info mation currently available from li A

amount required by the refineries depends on their gasoline output rela-tive to residual oil or bunker fuel, and while there is no indication that the the m. dustry as to expected future shipments (Tr. 3(M3-44,3078-79,3220-demand for gasoline, and thus butane, will be greater in the future than in 21). Unfortunately, projections up to 40 years are not available (ibid.).

the past several years, Staff nevertheless selected ten butane ships per year Intervenors also argue that using Staff's figure for butane instead of the as a conservative estimate of this parameter (Tr. 3216-17). We believe that Applicants', produces a risk from LPG which exceeds the risk from LNG Staff's estimate of ten butane tankers per year is preferable to Applicants, alone (Intenenors' Proposed Findings, paragraph 25a). The evidence estimate of two per year because the lower estimate leaves little margin c ntradicts this claim (see Table II, p. 676). Inter'venors have presented us with no alternatives to the estimates of Applicants aad Staff and ap-for market changes.

parently would have us simply speculate about future traffic. But our decision cannot be based on speculation. Consequently, we do not accept Butadiene j

i Intervenors' argument.

78. The Applicant indicated that butadiene is shipped up the Delaware l

81. We conclude, based on the above findings, that the estimate of a River to Mantua Creek on an average of ten times per year (Kalelkar Sup-f total of 60 LPG ships per year, consisting of 40 prepane tankers, ten "Dr. Read's response modifies his earlier testimony that the estimate of ten butanc butane tankers, and ten butadiene tankers is reasonable. We accept these tankers per year is based upon "present traffic"(Read's Supplemental Testimony at 9).

estimates in reaching our decision.

678 679

B. Accideitts Per Year u

E g

g g

g

82. The accident rate which we have accepted, supra, for LNG traffic, f

3

~

o i.

v.

1.51 x 104 occurences per mile-year, is also applicable to LPG traffic.

E Moreover, LPG tankers will transit under the same COTP rules that

[

govern LNG ships (Kaleikar Supplemental Testimony at 54-55). In addi-a E

9 2

tion, at least two-thirds of the LPG transits (the 40 propane shipiaents) n m

d 6

2 will involve modern vessels which are less collision prone than conven-3 y

j 2

m tional tankers and cargo ships (ibid.). Thus that the real accident rate for LPG S

ships, considered as a whole, will be less than the calculated rate. We con-

,c

., c L

clude, therefore, that the estimate of 1.51 x 104 accidents per mile-year is g

y conservative when applied to LPG traffic.

~

o go o

o m

n l

j AE C. Spills Per Collision c

83. The spill probability for L.PG ships was previously estimated by E

o 5

y g

l l

l l

the Applicants to be 0.02, based on the assumption that, because of their x

c H

small size, LPG ships are susceptible to collisions from all angles up to 90* (Applicants Exhibit 9, pp. 23-24). The Appeal Board asked why this E

value was not as great as the 0.05 probability detesmined for LNG ships 5

2 2

in the vicinity of the C & D Canal, where LNG carriers are susceptible

~

n

,h j

2 j

2 5

6 to collision angles up to 90'(ALAB-429,6 NRC at,244). The reason for y

g M

this difference was that LPG was being shipped in vessels outfitted with H

steel pressure vessel tanks at the time the previous analysis was made g

t g

g (Kalelkar Supplemental Testimony at 53). The tanks have greater rupture E

resistance than the refrigerated tanks used on LNG vessels, and

~

N 5

6 d

2

[

M t

consequently the spill probability was lower for LPG tankers than for I

A o;

LNG tankers (id, at $3-54). New Coast Guard regulations, however, will require that LPG ships be similar in design to LNG ships in the future; g

y e

g6 j6 those ships will also be required to comply with the same regulations in E

S g

9 4 4 4 4 2

transit as LNG ships (id., Appendix B; Tr. 3454-55). Therefore Appli-3 4

E.

Q 09 y

UM cants no longer use 0.02 as the spill probability for LPG.

x x

W 2

G 8" n Sm o

a A

0 m8 b

d6 66 Propane

~

5 vs

84. While the modern ships that will be used to transport propane will

.o

.E E

not be as large as LNG ships, they are larger than vessels previously used E%

M o

.E S

.g

.y.5

,y

}

for this purpose and will have essentially the same spill resistance as LNG g ",

[$

M 5-5E 2,

A tankers (Kalelkar Supplemental Testimony at 53-55; ailso see Appendix F).

.I

5l2 2y 5 $

8 Consequently, Applicants now believe that the probability of spill.'or e5 S$

f g ~j {

[g

[y h [S propane tankers should be the same as for LNG ships: 0.005 for the river section of interest, except in the vicinity of the C & D Canal, where it is i

681 680 l

I 0.05 (id. at $5-56). Staff is now in agreement with this analysis (Staff Pro.

l ability of 0.I, twice that of LNG carriers susceptible to 90* impact angles, In our view, these assumptions are very reasonable and, consequently, we posed Findings, paragraph 109).

find the Applicants' spill probabilities for LPG tankers acceptable.

Butane and Butadiene D. Vapor Clouds Per Spill

85. No information is available about ships that may carry butane on the Delaware River in the future, because there are no plans for future

87. The Appeal Board raised the same questians with regard to the shipments (Kalelkar Supplemental Testimony at 59-60). Consequently, the nonignition probability for LPG that it did for LNG (ALAB-429,6 NRC Applicants assumed that future shipments will be carried in vessels similar at 244). We considered this matter above, with regard to LNG (see para-to the Faraday, the last ship to carry butane on the Delaware (ibid.). On graphs 63 through 69). Based on those considerations, we concluded that the basis of this assumption Applicants used 0.1 as the spill probability the use of 0.1 as probability of nonignition, given as spill, was reasonable for butane, given a collision (id. at 60). With respect to butadiene, the and conservative. We make the same finding here, for the same reasons, tankers are expected to be small (id. at 59)." Consequently, a spill prob.

with regard to the probability of nonignition of LPG, given a spill.

abit;ty of 0.1 was assumed for butadiene. as well(ibid.).

ti6. The Joint Intervenors argue that Applicants have assumed that E. Meteorological Factor LPG ships will have double bottoms, and they suggest that " differences between the physical characteristics of LNG and other double-bottom gas

88. The Appeal Board noted that propane (LPG) b Wmmable at con.

vessels raise doubts about the applicability of the Applicants' analysis of centrations ranging from 2% to 6%, whereas methane (LNG) is flamma-LNG tankers to these other vessels" (Intervenors' Proposed Findings, ble at concentrations ranging from 5% to 15% (a 2.5-fold difference),

paragraph 18). To support their argument, they cite the testimony of and questioned why the meteorological factors of the two types of gas did Commander llenn, who in fact did say that not all LPG ships are double not reflect this difference (ALAB-429, 6 NRC at 244-245). Evidence pre-bottomed (Tr. 3464-65). Commander llenn went on to testify, however, sented in the remanded proceeding demonstrates that flammable limits for that LPG ships without double bottoms would have pressure vessel cargo gases, when expressed in percentages, are mole-percentages (Kaleikar Sup-tanks, not membrane tanks (ibid.). lie also said that pressure vessel cargo plemental Testimony at 56). In terms of molecular weight, propane is 2.75 tanks do not require a secondary barrier because "it's an overdesigned, times " heavier" than methane (ibid.). When the flammable limits of the very rugged tank" (Tr. 3455). When asked whether such ships would be two are converted from mole-percent to pounds per cubic feet, the lower any less safe than double. bottomed ships, he replied, " Absolutely not, llammable limits of the two are approximately the same: 2.59 x 105 lb/

sir" (Tr. 3466). Commander 11enn then provided detailed testimony ex-ft' for propane and 2.24 x 10 8 lb/ft8 for methane (id. at $7). The distance plaining why the Coast Guard considers pressure vessel tankers to be as that a vapor cloud remains flammabic is a direct function of the safe as double-bottomed tankers (Tr. 3467-69). From the evidence before flammable limit expressed in units of mass. Since in mass units the lower us, we conclude that LPG tankers equipped with pressure vessel cargo flammable limits of the gases are about the same, the maximum hazard tanks may be considered at least as safe in a collision or a grounding as distances for them are about the same (ibid.).

double-bottomed vessels equipped with membrane tanks. This finding is consistent with our earlier finding that freestanding tanks, i.e., pressure Propane vessel tanks, would be less susceptible to a spill than membrane tanks (see paragraph 57, supra). In any case, it is not true that Applicants apply the

89. In the earlier analysis, Applicants calculated the meteorological spill resistance of LNG ships to all LPG ships. Propane tankers were factors for all LPG on the basis of an assumed 10,000-ton spill (Appli-considered to have the same spill resistance as LNG ships, because of cants' Ex. 9, p. 50). The tankers which will be used for propane ship-their size and other structural and performance characteristics. Butane ments on the Delaware in the future, however, have tanks that hold an and lutadiene carriers, on the other hand, were assigned a spill prob-average of 9,000 mi, which is less than 10,000 tons (Kaleikar Supple-3 I

mental Testimony at 54). In the revised and updated analysis, Appli-38ne capaciiy or a typical butadiene tanker is 12.000 m, compared to 50.000 to 75.000 I

cants assumed that size of a propane spill would be 12,000 m', or 4,500 m for a propane tanker and I25.000 m' for an LNG tanker (id. at 54. 59).

3 682 683

tons (id. at 54,56). Under the most adverse weather conditions, a vapor F.

Probabilley of a Flammable LPG Vapor Cloud Reaching the cloud of propane from a spill of this size on water could travel a distance pg,,, ggg, of 9 miles and still contain flammable concentrations (id. at $4-55). On the basis of these facts and assumptions, Applicants calculated the

92. The probability that a flammable cloud of LPG vapor will reach meteorological factors fc: propane to be 0.307 for the entire river section the Hope Creek plant from an accident Rn the river is obtained from the of interest except in the vicinity of the C & D Canal, where it is 0.003.

d h p h M h M i d @ d min h M p d These values are less than comparable values obtained for LNG because LPG. These values are summarized in Table 111 (p. 680). The Applicants of th difference in spill size,12,000 mi for propane versus 25,000 m for obtained the following probabilities that a gas cloud would threaten the plant: for propane,1.0 x 108 (Kalelkar Supplemental Testimony at 55);

LNG, and the difference in flammability (/d. at 56). Previously, we con-for butane,1.0 x 10 8 (rounded up from 9.7 x 10') (id, at 60); for buta-cluded that the method used for calculating the meteorological factor was diene, 3.8 x 10-s (id. at 59). The cumulative probability for all types of reasonable, and the Appeal Board accepted that methodology. We find LPG based on the Applicants' calculations is 5.8 x 10 8 occurrences per here that the assumed 12,000 mi spill for propane is also reasonable. We year. For purposes of comparison, the estimate obtained by Staff in Sup-conclude, therefore, that the meteorological factors of 0.307 and 0.003 piement No. 5 to the SER is 6.4 x 10 ' occurrences per year (see Staff for the 24-mile river segment and the I-mile segment at the C & D Canal, Exhibit 1-F). We did not accept Staffs estimate of the number of respectively, are reasonable and acceptable.

propane tankers, however, and Staff did not consider butadiene. More-over, the 2 x 102 (0.02) spill probability is no longer acceptable. Con-Butane sequently, we reject Staff's estimate. We have found, in addition, that Applicants' estimate of number of butane tankers is unacceptable. We

90. In the absence of any plans by Gulf and Exxon to receive butane must, therefore, also reject the Applicants' estimate of cumulative pmb-at their Delaware River terminals in the future, Applicants assume that ability for LPG.

the last ship to carry butane on the Delaware was representative of butanc

93. We have fot nd Staff's estimate of number of butane tankers and carriers (Kaleikar Supplemental Testimony at 59-60). On this basis, it was Applicants' estimates for the other parameters all to be reasonable. Using assumed that a butane spill would release 10,000 mi of LPG (id, at 60).

Staff's value for number of butane tankers and Applicants' values for the With this assumption, Applicants found that a butane cloud could travel other parameters, then, we have calculated our own probability estimate 9 miles and still contain llammable concentrations. The meteorological for a butane cloud reaching the plant: 4.8 x 108 occurrences per year. The factor calculated on the basis of these assumptions and facts is 0.32. We values we used in reaching this estimate are summarized in Table II (p.

find the Applicants' assumptions to be reasonable and conclude that the 676). We have found the Applicants' values for the parameters associated meteorological factor for butane is also reasonable and acceptable, for the with propane and butadiene all to be reasonable. We conclude that the reasons set forth in the preceding paragraph.

probability estimates of 1.0 x 108 that a propane cloud will reach the plant per year and 3.8 x 108 that a butadiene cloud will reach the plant Butadiene per year are both reasonable. We have also found the collision rate and the nonignition probability (vapor clouds per spill) used in all the calcula-

91. Butadiene shipments are transported on the Delaware in a number tions to be conservative, for reasons discussed previously. Consequently, of different tankers, but the typical tanker has a capacity of 12,000 m' we find the probability estimates above to be conservative.

and a tarik size of 4,000 m8 (Kaleikar Supplemental Testimony at 59).

Based on its flammability limit of two mote-percent and a 4,000 mi spdl,

{

94. Accepting these probability values as reasonable and conservative, we have calculated a cumulative probability for LPG of 9.6 x 10s occur-Applicants de: ermined that a flammable cloud of Sutadiene produced in an accident on the Delaware River could travel about 4 m les (ibid.).

rences per year (see Table 11, p. 676). Considering the evidence before us, From the foregoing information, the meteorological factor for butadiene and for the reasons set forth above, we find this estimate of the prob-ability that a flammable cloud of LPG vapor, resulting from an accident was calculated to be 0.25 (ibid.). We find the assumptions with regard t nvolving a tanker on the Delaware River (except for a ramming of tower a butadiene spill to be reasonable, and therefore conclude that the 97), will reach the llope Creek site to be both reasonable and conservative meteorological factor of 0.25 is also reasonable and acceptable.

pursuant to the guidelines set forth in NUREG-75/087, $2.2.311975).

684 685

Captain Wiman's figure of 8.75 as the distance from tower 97 to the Ill. TOWER 97 llope Creek site. We conclude, therefore, that 8.8 nautical miles is the correct distance from the tower to the Want. Finally, tower 97 is located

95. There remains to be considered the possibility that an LNG or west of New Castle Range, about 800 fert from the edge of the shipping LPG cloud will reach the liope Creek site from a ramming accident in-channel (letter from Captain Wiman dated November 10,1977; Tr. 3738, volving a tanker at tower 97. The circumstances which called this matter 3746).o to our attention and caused it to become of concern to ut have been

97. The tower is constructed of tubular steel and is about 380 feet tall described supnt, in paragraphs 16 and 17. We turn now to a considera-(Tr. 3756). It is anchored to a couret: platform, the top of which is tion of the evidence on this subject, received by us in the haring on about 14 feet above mean water level (Tr. 4755). There is a fender system January 10,1978.n of concrete pilings surrounding the tower on the east and east-southeast

96. Tower 97 is being constructed upstream from the Ifope Creek site, sides, between the tower and the channel (Tr. 37%, Applicants' Ex.16).

near the north end of New Castle Range (letter to Chairman Luton from The fender system is designed to withstand a ramming by a 2,000-ton Captain K. G. Wiman, U. S. Coast Guard, dated November 10,1977).e barge striking it with a velocity of 3 or 4 feet per second (Tr. 3756-57).

The exact distance from the tower to the flope Creek site is not made Navigational warning devices which will be installed on tower 97 will con-clear by evidence which we have received. Captain Wiman informed us sist of flood lights, to illuminate the base, and a foghorn that will be that the " distance from the towers to flope Creek Station site on Artifi.

audible for one-fourth mile (letter from Captain Wiman dated November cial Island is approximately 8.75 miles" (letter to Chairman Luton from 10.1977; Tr. 3770-71)." In addition, the tower will be equipped with red Captain Wiman, dated November 29, 1977)." Dr. Kaletkar, on the other flasliing lights to warn aircraft (Tr. 3738, 3772). Tne water at tower 97, hand, testified that tower 97 was 9.1 nautical mdes from the liope Creek according to Captain Wiman who obtained his information from the site (Tr. 3748). Since this inconsistency could not be resolved during the Army Corps of Engineers, is 34 feet deep at mean low water (letter from hearing, the Board asked LTJG Stanton to have another letter sent to m Captain Wiman dated December 19, 1977)." Applicants' witi,c.: Boettger, by the Coast Guard informing us of the correct distance from tower 97 to who inspected the tower from a fisheries research vessel on Jar uary 6, the llope Creek site (Tr. 3785). On January 27, 1978, we received a letter 1978, testified that the depth finder on the boat registered depths varying from Captain J. C. Griggs, U. S. Coast Guard, advising us that he had determined that the distance from tower 97 to the " southern dome of the l

"from something like 29 feet to 32 feet" at about mean water; these depth readings were t'aken at distances of from 100 to 200 feet from the existing Salem generating station," the closest reference point on the I wer (Tr. 3739-41, 3745). The deeper reading was around the northeast chart, is 9.1 nautical miles.*2 From information in the flope Creek FES, corner of the tower, and the more shallow reading was around the south-we have determined that the southern dome of the Salem plant is about west corner (Tr. 3739). Witness Boettger disclaimed any personal knowl-1,950 feet or 0.32 nautical miles, from the northern dome of the llope edge or expertise about the accuracy of the depth finder on the boat, but Creek plant (see Final Environmental Statement, liope Creek Generating he indicated that he understood from others who possessed such knowl-Station, Fig. 3.2). Using Captain Griggs' information, then, we obtain 8.8 edge that it was accurate to within il foot (Tr. 3739, 3742-45). While nautical miles (rounded to one decimal place) as the distance from tower witness Boettger's testimony cannot be accorded the weight that could be 97 to the llope Creek plant. We consider this value to be consistent with accorded testimoity from an expert who is reporting results from careful and deliberate measarements, the Board i evertheless notes that Mr.

" Evidence at the hearing was presented by witnesses for Applicants,!.taff, and the Board.

Boettger's observation that deeper water wa recorded northeast of the Joint Intervenors presented no evidence but relied on cross-examirution of the witnesses. In tower and shallow water was recorded southwest of the tower is consistent addition, the parties stipulated to submit the record as it stood at adjournment of the with, formahon contained on the photocopy of a section of NOS chart-m hearing. Lt., withoe their submitting proposed findings of fact and conclusions of law (Tr.

12311 which was attached to one of Captain Wiman's letters (letter 38uu2).

"I'he correspondence between Captain Wiman and the Board was received into evidence at the hearing on January 10,1978 (Tr. 3748).

OSee p. 686, n. 40.

• Seen.40.

• • See p. 686. p. 40.

All parties were aware that this information was soticited by the Board and none

1. See p. 686, n. 40.

42 objected to it (T r. 3715).

687 686

sake of conservatism, that tower 97 is one of 50 rammable objects on the from Captain Wiman dated November 10, 1977).** Th-u chart shows a river which might pose a risk to the Hope Creek plant (ibid.). This depth of 35 feet northeast of the tower, near the channel, and depths el assumption yielded a probability of 2 x 108 that, given a ramming in the 26 feet in two places southeast of the tower, toward the channci (ibid.;

river section of interest, the object rammed would be tower 97 (ibid.;

Tr. 3fil5). All of this evidence leads us to believe that the only direction an LNG tanker could approach the tower and ram it would be from the Applicants' Exhibit 14). Taking the probability that a large ship will be involved in a ramming accident,1.9 x 104, as the probability that an LNG northeast. Loaded LNG and LPG tankers traveling to terminals for un-ship will be involved in a ramming, and multiplying this by the probability toading will be approaching the tower, in normal operation, from the that tower 97 will be rammed, given a ramming, gives a probability of 3.8 south-southeast. We believe an LNG or large LPG vessel would run x 10* per trip as the probability that an LNG ship will ram tower 97.

aground before ramming the tower from this direction. Nevertheless, in 100. Stalf's approach to estimating the probability that an LNG ship making their analysis of the probability of a tanker ramming tower 97, would ram tower 97 differed significantly from Applicants' approach (see the Applicants conservatively assumed that the water around the tower Read Tower Testimony). First, Staff used its previous estimate of 360 ships per was more than 35 feet deep at all times (Tr. 3768). This assumption would year (Read Tower Testimony at 4). Then Sraff obtained a data base of 26 adequately account for smaller LPG (ankers which might be able to rammings in an 8-year period, 1967-1974 (id. at 2). The difference between impact the tower by approaching through water which is less than 35 feet Staff's figure of 26 rammings in 8 years and Applicants' figure of 15 deep.

rammings in 8 years may result, in part, from a large number of rammings in the years 1967 and 1968, which were included in Staff's analysis but A. LNG Traffic not in the Applicants' (Tr. 3834). The difference can also be attributed to the fact that the Applicants considered only rammings which involved a

98. In their analysis of the problem of a possible ramming of tower 97 vessel having a draft of over 18 feet, while Staff's data included small by an LNG ship, the Applicants used 292 as the estimate of the number vessels such as barges (Tr. 3830). Also, Staff's data included rammings of of LNG ships that would pass the tower each year. This is the same number used for LNG traffic in the collision analysis (Tr. 3748-49). The stationary vessels, while Applicants' did not (Tr. 3837). Staff's analytical approach involved a geometric model in which it was' assumed that tower rate of rammings was estimated on the basis t:f historical information on 97 obstructs one percent of the river, and that ships proceed up the river rammings on the Delaware, obtained from Coast Guard accident casually at random, without regard to the channel and without their pilots detecting data for 1969-1976. An analysis was made of all rammings involving ships the tower until the ship arrives at it, at which time the ship may or may of more than 18 feet draft (ibid.). Even though these accidents included not avoid it (Read Tower Testimony at 2-3). This model predicts the seven occurring at night and one occurring at ancht, rage, all of which are occurrence of I x 105 rammings per ship passage (id. at 4). This risk was accidents that would not be relevant to an LNG tanker in the 24-mile catchment zone, for the sake of conservatism Applicants included all 15 l

applied to LNG tankers by accounting for their size,600 meters in length, in their cakulations (Tr. 3479, 3810-II). There was an average of 9,500 calculating a risk per meter, and then prorating that risk out to I mile (Tr. 3833). The result is a ramming rate of 2.7 x 104 per mile which Staff one-way trips per year during the 8-year period from which the data were takes as the effective ramming rate at the tower (Read Tower Testimony taken, to give an estimated probability of a ramming accident involving a at 4).

large ship of 1.9 x 10' per trip (Applicants' Exhibit 14; Tr. 3749).

101. Staff's estimate of the rate at which LNG ships would ram tower

99. To provide a basis for estimating the probability that an LNG ship would ram tower 97, given the occurrence of a ramming by an LNG ship, 97 is about one order of magnitude greater than the' probability obtained by the Applicants,3.8 x 10*. While it is very difficult to compare the two the Applicants counted 75 rammable objects in the river from its mouth estimates because of the difference in approaches, we do note that Staff to the Conrail Bridge near Philadelphia (Tr. 3750). Only one of these considered all rammings, not just those involving large vessels. We note, objects, viz., tower 97, is located in the 24-mile catchment zone (ibid.).

Most of the rammable objects are in the heavy industrial area north of also, that the geometric model on which Staff based its analysis is extremely conservative. In fact, in our view the model is so unrealistic Wilmington, and the 15 rammings accepted for the data base occurred in that we consider Staff's estimate unacceptable. Clearly the assumptions that section of the river (ibid.). Nevertheless, Applicants assumed, for the on which the model is based are invalid. We consider the Applicants' esti-i

• • Sce p. 686, n. 40.

689 688

mate of accident rate at the tower, on the other hend, to be reasonable.

additional energy would be absorbed by the first tank to be struck (Tr.

We believe the historical data base is sound and the accident screening is 3805, 3817). Only if it fell horizontally, which is extremely unlikely, justified. Further, we find the estimate to be conservative because the 15 would it strike more than one tank simultaneously (ibid.). In that case the rammings included accidents which 9ccurred under circumstances that energy would,be equally divided between the impacted tanks (ibid.).

are not applicable to LNG tankers, ad because the number of rammable Consequently, it is highly unlikely that a two-tank spill would occur; wit-objects (50) used in Applicants' analysis is substantially greater than the ness Kale!kar testified that the probability of a two-tank spill, given the real number (1) which an LNG tanker could ram in the 24. mile catch-ramming of tower 97 by an LNG or LPG tanker, would be on the order of I x 10' or I x 10 m (Tr. 3806). In view of the evidence, we conclude ment zone.

102. Next, we must consider the probability that a cargo spill will that the probability of a two-tank spillis negligible.**

occur, given a ramming of tower 97 by an LNG tanker. If an LNG tanker 104. With regard to the probability of a one-tank spill, Applicants wit-rammed tower 97, a cargo spill could occur by one of two general mech-ness Athens testified that in view of the many factors involved in de:er-anisms: tank penetration or rupture resulting from hull damage, or tank mining whether the tower falls and, if so, where it falls, he would estimate penetration or cracking resulting from the tower falling onto the ship (Tr.

the probability of a spill given a ramming to be somewhere between 10' 3758-62). Because of the geometry of the fender protection system rela-and 102 (Tr. 3761-62). As a conservative estimate of this probability, tive to the tower and to the channel, the likelihood that an LNG ship therefore, Applicants have used the value of I x 10 8 as the probability of could strike the tower without first ramming the fender is negligible (see an LNG spill given a ramming of tower 97 (Applicants' Ex.14). We Applicants' Ex.16 and the photocopy for NOS chart 12311 attached to recognize that this estimate is based on engineering judgment and is not a the letter to Chairman Luton from Captain Wiman dated November 10, quantitative evaluation. We also recognize that the number and complex-1977)." Since the fender system is not designed to withstand impact from ity of the parameters which must be considered in making this determina-a large vessel, it would probably be pushed over or pushed aside if struck tion make a mathematical analysis extremely difficult, if not impossible.

by an LNG tanker (Tr. 3756, 3758). The vessel might then proceed to the Therefore, we will accept an estimate based on engineering judgment.

tower and impact the base, or both the legs and the base (Tr. 3759), Con-Without explanation, Staff used the same probability of spill as in its sidering the structural resistance of the ship, relative to that of the fender collision analysis,1 x 10' (Read Tower Testimony at 4). Thus, Applicants and the tower,it is unlikely that such an accident would cause significant and Staff are in agreement as to the value to be used here. Considering damage to the ship (ibid.) Consequently, if the impact did not cause the the evidence, we conclude that the Applicants are correct in estimating the tower to fall on the ship, we do not believe that damage to the vessel probability of spill, given a ramming, to be less than I x 108. We accept ramming the fender or tower, or both, would be sufficient to cause a that value, therefore, and find it to be conservative, 105. We turn now to a consideration of the probability of nonignition cargo spill.

103. If the impact of the ship were sufficient to cause the tower to fall, of an LNG gas cloud released as the result of a ramming of tower 97 by one of several things could happen. The tower might be pushed forward an LNG tanker. Applicants and Staff both used the same probability (1 x or to the side, so that it fell away from the ship (Tr. 3761, 3798). For the 10') in this analysis as they did in the analysis involving ship collisions reasons just given regarding structural resistance, we do not think that (Applicants' Ex.14; Read Tower Testimony at 4; Tr. 3768). In response such event would cause a cargo spill. If the vessel impacted the tower with to questions by the Board, Applicants' witness Boettger testified that the sufficient force to break the tower away from its foundation, however, transmission line to be supported by tower 97 would be carrying 500,000 inertia effects might cause the tower to fall back on the ship (ibid.).

volts; witness Kalelkar said that this fact was not taken into account in using a nonignition probability of I x 101 (Tr. 3768). Mr. Boettgo testified Should that happen, at least one of the cargo tanks could be penetrated or cracked, resulting in the release of LNG (ibid.) Since the tower is 380 feet further that sparking would occur if the cable carrying current parted or tall, it is v.uccivable that it could strike two tanks (Tr. 3798). As the came in contact with the tower, any other conductor, or the water; this sparking would last less than a second before power in the cable was shut tower fell, however, energy would be absorbed as it impacted the bow or i

side of the ship; it would strike one tank before striking a second, and off automatically (Tr. 3823). We have concluded, supra, that the only asWe consider a three-tank spill incredible because cargo tank f 3 on an LNG tanker is osee p. 686. n. 40 more than 380 feet from the bow (see Board Ex. 2 at 173).

690 691

credible scenario which would produce an LNG spill if a tanker rammed TABLE IV tower 97 involves the tower falling onto the ship and penetrating or crack-ing a tank. Given that situation, it is likely that the cable carrying current Applicant's Estimates of Parameter Values and Resultant Probabilities for would fall and produce sparks on or near the ship, if the cables did not LNG Traffic and LPG Traffic at Tower 97 produce sparks, we believe that the nonignition probability, given a gas cloud produced as a result of the ramming of tower 97 by an LNG ship, LNG LPG

• would be about the same as the nomsnition probability given a gas cloud production in a ship-to-ship collision. We have accepted I x 108 as a reasonable and conservative probability in that situation. We also accept Tanker Trips 292 52 that probability here for the same reasons, but we consider it even more per Year conservative in this situation because of the likelihood that the cloud would be ignited by sparks produced by the transmission lines.

Probability of l.9 x 104 1.9 x 104 106. The last parameter to be consider in the analysis of the risk posed Ramming per Tr.ip by the possibility that an LNG ship might ram tower 97 is the meteorology factor. Again, we note that the Appeal Board accepted Ap.

Probability of 0.02 0.02 plicants' analysis of the meteorology factor (ALAB-429,6 NRC at 242).

Ramming Tower 97 Applicants and Staff both estimate that the probability that a methane gas cloud released by a ramming accident at tower 97 will reach the Hope Probability 0.1 0.1 Creek plant in Hammable concentrations is 2 x 105.* We accept this of Spill value.

107. Based on an estimated 292 'ransits per year by LNG tankers and Probability of 0.1 0.1 on the foregoing probabilities, App' cants calculated a probability of 2.2 x Nomgmtion 108 that the llope Creek plant would be enveloped by a flammable gas cloud resulting from the ramming of tower 97 by an LNG ship (Appli.

Meteorological 0.002 0.002 cants' Ex.14). We have found the number of transits and the foregoing Factor probabilities estimated by Applicants to be reasonable. We therefore find the cumulative probability to be reasonable. We have also found the Product 2.2 x 10 s 3,9 x 10 e conditional probabilities used to obtain this product to be conservative, for reasons already explained. Consequently we find the estimate,2.2 x

• The Applicant did not calculate a probability for LPG traffic at tower 97.

10 8 occurrences per year, to be conservative (see Table IV, p. 693). We The parameter values listed are from Applicant's analysis. The probability have rejected Staff's method for estimating the probability related to was calculated from them by the Board.

tower 97 on the grounds that it required unrealistic assumptions. But we did find, previously, that Staff's estimate of 360 transits per year was reaunable and more conservative than Applicants' estimate of 292. Using Applicants' method with Staff's estimate of number of transits, we have B. LPG Traffic calculated a most conservative estimate of 2.7 x 108 occurrences per year (see Table V, p. 694). Our decision is based on this value.

108. The Applicants did not consider an LPG spill at tower 97 to constitute a risk to the plant (Tr. 3749). Although the distance that an LPG cloud will travel, while still remaining flammable, is about the same "This estimate is based on the belief that the distance from the tower to the site is 9.I as the distance that an LNG cloud will travel and remain Dammable, miles. We have found, supra. that this distance is actually 8.8 miles. The small difference in distance does not affect the value of the meteorological factor (see Applicants' Ex. II, p.

LPG w 11 be carried in much smaller cargo tanks than LNG (Tr. 3769).

2a).

The average tank size of LPG ships that will be serving the Sun Oil 692 693

- d aae.

a l,

s g,ecef

.)

r r t a - - n ns e sdGotab ipe s e cNimGr ngl ea n e wf lb a( v wn nmis di lc rRp inP ed, )

t a

eh u

r i

orI e e o i a s

i t

s t e r uP o

h o aoh a

s n

nt o o L yC r sPpgi v f

s s

pdnTi age i

,LspiLsf meh a i

l r

v c

t c rTn ye r o e s e

mi e e aoe

.ht i r s

t d o e

r a r r f

f a pomwit ih ue i n v

t f

osNe r

r n nif s

r r 5 n e mc ah y aga e or o v ei aoh r d

o e u pa s

h o is z i

a a1 c

rh hd rh wo l

r t

G ouot 5 lck a a u a n,i sh ipk tyla e gm gGce6t e

t ms m 'o o h mact nh t

eit s

r cf e et s u na

.d s

s ee ocl nd t

or

,f r

a s

(

pe d' leNe

,f mn o2 e e a

b n i l e

u phh r e e f r r

n9 o a nc e

t r e r r t

b L 2

e niFi c

,d r

l n

el e aa y s ws oil lc t

t i

s a

s4 d v t

Na c l c

a e x - n amy ay lpl c r s ab b c t

t s

c o

,e m

nr e o1 r ia e ot o

ni v

t i

t e

t f r cl e

Li a v Bma Bu a

t. bt o

t i r r a ie e g a

at e m gi A s in( a gf o oib p

f ts t

r s a. c et x

o ol e t

age e s f

pi eh kbjh eb o inf mbG c e c mnc Y g an rLe h sE g mPe h pt e ab t loc nk a a a

t m oo oP it a c nl s

e n

r ua T if i eAh lot e sf i r

f mLv sTs r i e a t

t v

s n nof e er gogh pL h e o e aGl i ip

. c aCo r

h(

c s

s lot g

i s

t a ei pab et f

s I

r e a my e

l a

t s

r e L n a e e l

t mo aS s f nah pl f t

icP av s i nme b

ai gne d h u

e v

mh t

t l s

pW cf nh t ipai i

I la e o nt eGl oami e b. kinoh y

r s e pl an i

i t

it rl Loiwfio ihGosnoh B

mr a gLa r

y it i

r r

ws c

o r o f

e t

l e.

A r s e

ini ibi5 B e y ngdf dGe m,h

.mlai n

r e, ol f o nPa ue t

t s

l mns 4a r

v) et a

n c l r sFooh a

g u

e w

oLa c eee ah b pid at sh ek e

t e n teP t i di od ne b a7 Ohi ici cn iaL 3 d r

r t lt g a1 mn.oe r t c

l dl ad - r e

u 0 n ol hh r b

w a p )s os o v9 R

b c

5 e

e d

a et '

t t

s e6 t

r1 t n a

e net mt a0 e r io' a

0 ps(

P h ac c ia s n r

od o 2 z 4 n u t

e s

a e

s 0 nn n5 mahGlumim1 t a e(

f b ae s

a h

e r it

. a n

r o at s a oe a nd oi n3 E wo n gb a

d ph o e, p s a sT.No l

1 t

x am loxa et r pi 0t c h c2 V

a na e

ah el t

',h o

e o r

o sGd t

5 r

r s

r t

ap sLvGe9 u

a 9

I d p k Net, g t

i ll sAa c e

e e) 1 t

r emi na o2 T e s

ph n 6

6 iLWc9.le t s r

r weh(

e i N 's 3 a h e 3

if me r

u si e

ht x

d s6. A s os plt f e) t t

r n

r v o r

i

. ih4 at n a

,o L

gi r in t

h

).

xb e

e y

a t

ts rl l

t f

us l

a et n

l n

,1 oa h

t t f d i

i7 Ea t 3 s f

a oi9 eno oi s f ic f

7i aa i at h

si e r oaa i

s es gl t w2 e m

t cd e r n d 8i U

lut nabi d g, ta 't m

r e a w gh s

l c

1 k r ue nhi y 6

t i

e ad g s

e n0 c e

e Md t

t i r t t d kh eol w

t a e cb at lpl k

.h!

c s

miot a e n6 na s

a/ e i e pt b ah ei a e r

s n

l h bl u o2 n e a r ut u

,l 7

o uk e

gl 1 i af nt t

a c cb 5d U

a, s y pi ed no e

mcGkl k

t c

b e r

,ga imt t

a s r

r r a, si e o

t nGPnpAabCVntpt pt at t

c e o C B c et t

a 4 'ylph hhl s

o nb u

e u e s or d

i Am t

t a s

o aGo e

e r ur p e e0t pt ape i e e e m

ngl ct is g

o P Lt a s o I

s l

v r1, ic ai y c e nE V ahCmp t

a e l

r e uc oLrGl u p pl c

cGid Nl am u'soRn nr e

t e

oa h d9 a e o I

e y c t

e e s lat l

ob yat L

ps pet h i 3 f f

o r

ef Pl e al b

i aPs s

ef wtph eLh w a

el Le msun uoe c aUi e

t s ns s

a i e s

e h

pi owt T r

e(

oh h c g t

t nh nrNc At H n

l u aa oc t

s n

Tdb b n

c pt d inni nis cc t

o a

,e gn fi e e

r e n o aWah e

i hla H.

mb a

r d

at l t

as

.c e s

i a) e b n et ec p) e r oat t

c o aeia h b

gt zi e s ie la t

r t

s xt h w

.b s

t ei ad o

).

y h pei w5 n o f

s i

h ahi e

a h e8 t d rk ad e t

t r

1 r

b es

( a g pt m.Ve gwh.

t h s, eb h

c d o Th F

,d o noh(

t ni e

t inl as t

t t

upi e o I t r

,o b np4 vt a

e niAig sd r

o le r r e o e ri ec7 o u 97 i am e

7i a p r

n n l

a9 oa69l t

)

pi i at 0 ins gn n

i s ai a sd 9l getaw nir n n

r e

ef e c 9

per vl s

o a eu

,r rb s. rbih2 oml

,e1 l t

p I

e e a l

s inf md ih e

g 0 r pb l wc l

c( r r 1 d cki J

w l

e l

e ed pe rd e

al r

pi pmnn tndloman e

b nr 1

b wah cy wt ch yVt,oWAw9 paiioov a o u n ans3 H

o ei r o l al wnvl w e ph z

a ei oaai 7

e a e4 e in e

oa a a ar st o Tc p f r t d2 mF pf h m l

t r

j r

ae e e o 8. bet k c

d o

nbh n e.s n

f d a t

8 i

r r u

e po d sl r o

l d

s t

r u o meii oe r

t t s ch y

s a

r eo oi t

b s pc r o

e t

nf pt d et n B

0 naa al h a i n

o d

2 e

e cGh ut ht i

G 2

1 a

t e

0 0

1 M

o I

ct s t

t a t

h P

6 0

E ainP wf a

x i

hd aLd r

l t

L 0

os e b

b R

6 pt t

e e

nt h ml inu id o

y p

s t

t r

n i gi eai t

4 s

Achd ai s

a e

n w

q d

. ad et rid e o e

)

s h

eh nr c ts nt dt t

1 n

eWi p

0i e

t o

f ma it u

e 8d c

ot dd

. r sh a

3 a u c

ef hl e r uit lu A

.h dt t

c ve oah lc r

nav vt a

t s

T nf e

( e oh e i

c f r i

t ae it s a

,li c w o e w n

e s

t cb n

woo io o

7 rl e e sWh nw i

V i9 e ul b l

s t

b 0

l e c n lo)

E a

r 2

1 el a ea. ei c1 4

e G

0 2

w N

6 0

1 ma mh ic ).

t v

L b

I rd 0

9 x

c di o

e8 t

B o

c omf lpi s

e h3 6

3 0

0 r o L

9 7

i t

a o pb inf e t

A PT i

c t

A( n d

nr T

2 b

o iI l

n a c

is a ne usf r

t t r

yl e e d(

si pv np t

e e t i

~

0 t ins e

ac ael e

e ue s r e abl ic e e

h l f imfah,hW led 00,m m uf s

s a e e Gt t

e e r 1 k oPtad.

imt 9 c rh RT ib d r

t o

e 9 p et Lh t

n9 dG uu n r

e r t

e f e nP oc aOs e e is t

l r

c r a

7 b 0l iCwr ed s

e e L

o v s

ip 9

a0 e S. tneo nil o its e n a i v

m m ed r

r e 0, a ).lapt oeige un T

e b2 c

6 rdpl b

a f

f w f

l t

l a

a c

1 i

h e

e8 t

r l

s o e oo o

ig d

iwolu(

c u

l o eb

's h n

l f d b e c

Vi ip p

yT y

y t

e i

t h e o

o o

t h m w

t s

f ea Tr ity

,G rf r

t n

t t

e g

g l

r eos a m eT a

l n

il n i

il it l

P o

ot eP ic t

l t r ibi ibi ibl b n r

a z e

t of cL lpf r

ii o

niwl s

i r r a o

l r

o d mrf mn m

e e am am ai a g o

imk u et am Au r

p aG k Y b

p e

si p

bi e

b n b

m n

t t

r o-oS o

r n

o r no6 wkl ig idf i

t o

e c aN a e r a r a rf r

s o si a

e al g

oiml PL Tp PR PR Po PN MF e

r n

t c8 t

t r

e

abilities for the mutually exclusive extents of (1) a collision involving an per year (see Table VI, p. 697). We find, further, that this calculation is LNG tanker, (2) a collision involving an LPG tanker, (3) the ramming of conservative pursuant to the guidelines contained in NUREG-75/087, tower 97 by an LNG tanker, and (4) the ramming of tower 97 by an LPG 02.2.3 (1975), because of the qualitative arguments set forp herein which tanker. We previously accepted as the criterion for our decision the guide-show that the realistic probability, although unknown, is less then the line probability values set forth in the NUREG-75/087, 62.2.3 (1975).

estimate which we have calculated.

Thus, an event resulting from the presence of hazardous materials in the vicinity of the plant may be disregarded if a " realistic" calculation of the TABLE VI event's probability of occurrence is less than 10' per year or if a " con

• Probability Estimates Accepted by the Board servative" calculation indicates that the probability is less than 10* per year (LPB-77-22 at 709-10). The event referred to is one which has the Risk from:

potential of causing radiation exposures in excess of the guidelines con-tained in 10 CFR Part 100. The Appeal Board also accepted these LNG Traffic Over 24-Mile Range 1.1 x 10' guidelines (ALAB-429,6 NRC at 234). Since valid statistical data are not available on the casualty experience of LNG and LPG ships in inland LPG Traffic Over 24-Mile Range 9.6 x 10 8 waterways, the estimates with which we must deal are based in part on data from analogous experience and in part on engineering judgment.

LNG Traffic at Tower 97 2.7 x 10 '

Consequently, the calculation of cumulative probability on which we must base our decision will not be " realistic." We can, however, consider it LPG Traffic at Tower 97 4.6 x 10' acceptably " conservative" if the probability is approximately 10* and is combined with " reasonable qualitative arguments" which show the realis-Cumulative Probability 2.4 x 10 7 tic probability to be less than 106 (NUREG-75/087,52.2.3).

112. The probability estimates which we have accepted above as being reasonable and conservative are listed in Table VI (p. 697). The prob-V. VINYL CIILORIDE THAFFIC ability that LNG traffic over the 24-mile range of the catchment zone will result in a flammable gas cloud reaching the liope Creek site,1.1 x 10' 114. Although. vinyl chloride is not included in the Coast Ger.rd's occurrences per year, is conservative because of conservatisms associated classification of LPG, the Applicants considered the risk to the flope with the estimates of collision rate, probability of spill, and probability of Creek plant from shipments of vinyl chloride on the Delaware !!iver nonignition. The probability that LPG traffic will lead to a gas cloud at (Kalelkar Supplemental Testimony a t 57, Tr. 2993). Within the past 2 the plant site,9.6 x 108 occurrences per year, is conservative because of years shipments of vinyl chloride have been received by PPG Industries conservatisms in the estimates of collision rate and probability of at Paulsboro, New Jersey, and it is expected that these shipments will nonignition. The probability that the ramming of tower 97 by an LNG continue at the current rate into the forsceable future (Tr. 3N3-44). There tanker would produce a gas cloud that would reach the plant,2.7 x los is an average of 25 tanker shipments of vinyl chloride per year, all of occurrences per year, has been found conservative because of conserva.

which are carried by one tanker, the Puerto Rican (Kalelker Supplemental tisms in the estimates of the probability that an LNG ship would ram the Testimony at 57). The Puerto Rican is a double-bottomed American flag-tower and the probability of nonignition of the gas, given a ramming and a ship having a draft of 11 m (about 36 feet), length of 200 m (about 656 spill. Finally, the probability that an LPG tanker would ram the tower and feet), and beam of 27 m (about 89 feet) (id, at $8). The tanks in which produce a vapor cloud that would threaten the plant,4.6 x 10* occurrences vinyl chloride is carried are arranged longitudinally in the center of the per year, is conservative because of conservatisms in the estimates of the ship and are surrounded by cofferdams. Between the cofferdams and the probability of ramming, the probability of nonignition, and the hull are other tanks used for nonflammable materials (id. at 57-58; Tr.

meteorological factor.

3N9). At their closest point to the hull of the ship, the vinyl chloride 113. In conclusion, we find the cumulative probability of a flammable tanks are 26 feet away (Tr. 3N6). Tankers carrying vinyl chloride are gas cloud reaching the llope Creek plant because of an accident involving subject to the same Coast Guard regulations as ships carrying LNG and an LNG or LPG tanker on the Delaware River to be 2.4 x 104 occurrences

. LPG (Tr. 3450).

696 697

' 115. Because of the size and construction of the ship, and because of expected to occur with probabilities less than 1 x 106, based on a the fact that it must transit the Delaware under the same Coast Guard conservative calculation, may be disregarded in the design basis of a facil-regulations as LNG ships, Applicants applied their LNG analysis to the ity. We therefore conclude, as stated in our order stated January 26, vinyl chloride tanker (Tr. 3046-48). Thus, the collision rate was estimated 1978, that the Hope Creek Generating Station, Units I and 2, need not be to be 1.51 x 10* (Kaleikar Supplemental Test l mony at 58). Conditional designed so as to protect against flammable t,as cloud accidents. We also probability of a spill, given a collision, was estimated to be 0.01 (ibid.).

reiterate the conclusion previously stated in our Supplemental Initial Deci-This estimate takes into account the fact that the tanks are 26 feet from sion of March 28,1977, namely, that the environmental impacts of LNG the hull and are surrounded by other tanks of nonnammable materials or LPG tanker accidents which might affect the plant are so remote and and cofferdams (Tr. 3046, 3097-98). A meteorological factor of 0.24 was speculative that there is no need to prepare and circulate a supplemental calculated for vinyl chloride, based on a tank size of 2,000 tons and the environmentalimpact statement covering this matter.

flammability characteristics of the commodity (Kalelkar Supplemental Testimony at $8-59). From these values, the Applicants calculated that the VII. ORDER probability of a Hammable cloud of vinyl chloride posing a threat to the llope Creek plant was 0.9 x 108 occurrences per year. The Staff has 118. It is ordered, in accordance with Sections 2.760, 2.762, 2.7(A, accepted this estimate (Staff Proposed Findings, 5129). The Joint 2.786 of the Commission's Rules of Practice, that this Second Supple-Intervenors direct the same argument at Applicants' estimate for vinyl mental Initial Decision shall become effective immediately and shall chloride that they did for LPG, vir., the Applicants have not adequately constitute, with respect to the matters covered herein, the final action of accounted for future increases in shipments of the commodity (Inter-the Commission forty-five (45) days after issuance hereof, subject to any venors' Proposed Findings, paragraph 25d).

review pursuant to the Commission's Rules of Practice. Exceptions to this 116. We find, with regard to intervenors' arguments, the same as we Second Supplemental Initial Decision may be filed by any party within did with regard to Intervenors' arguments concerning Applicants' seven (7) days after service of this decision. Within fifteen (15) days there-estmates of LPG traffic, and we reject them. We believe that the Appli-after (20 days in the case of the Regulatory Staff) any party filing such cants' estimates of vinyl chloride traffic, collision probability, probability exceptions shall file a brief in support thereof. Within fifteen (15) days of of spill, nonignition probability, and meteorological factor all to be the filing of the brief of the appellant (20 days in the case of the Regula-reasonable. We find further the probability that a vinyl chloride cloud tory Staft) any other party may file a brief in support of, or in opposi-will reach the plant in namn able concentrations is a conservative calcula-tion to, the exceptions.

tion because of conservatisms inherent in the estimates of chance of colli-sion and of nonignition. Although the probability that vinyl chloride ship-ments will pose a threat to the plant has not been included in our cumula-THE ATOMIC SAFETY AND tive probability for LNG and LPG, 2.4 x 10' occurrences per year, we LICENSING BOARD note that if it is included, the cumulative probability increases to only 2.5 x 10'. This effect is insignificant, and we conclude therefore that the threat from vinyl chloride shipments is negligible.

Ernest E.11i11, Member VI. CONCI.USIONS Oscar 11. Paris, Member l17. On the basis of the evidence before us, and for the foregoing Edward Luton, Chairman reasons, we have found that a conservative calculation of the probability j

Dated at Bethesda, Maryland, that a Hammable gas cloud resulting from an accident involving an LNG this 13th day of April 1978.

or 1.PG tanker could reach the llope Creek plant is 2.4 x 10' occurrences per year. This value is less than I x 10*, the guideline probability for a conservative calculation set forth in NUREG-75/087. Events which are 698 699