SIMULATIO OF THE VESSEL TRAFFIC SCHEDULE I THE STRAIT OF ISTAUL Şiri Özlem (a, İlha Or (a, irur Özbaş (b (a oğaziçi Uiversity, Istabul (b Rutgers, The State Uiversity of ew Jersey (a siriozlem@yahoo.com, or@bou.eu.tr, (b birur@ozbas.com.tr ASTRACT I this stuy, a simulatio moel is eveloe via the Area 11.0 software to mimic the actual Istabul Strait vessel flow uer the establishe traffic regulatios a meteorological coitios. The establishe ractice of ui-irectioal aytime a two-irectioal ighttime traffic scheules are reflecte a ilot a tugboats services scheule i the traffic flow irectio, visibility, curret a storm iformatio are also itegrate ito the moel. The effects of factors such as ursuit istace, vessel rofile, ilot availability, arrival rate a visibility over selecte erformace measures are ivestigate through sceario aalysis a the most imortat factors are etermie as arrival rate of vessels a visibility. Keywors: Strait of Istabul, Maritime traffic, Simulatio 1. ITRODUCTIO The Istabul Strait, 31 kilometers i legth is oe of the arrowest waterways i the worl with oly 660 meters at its arrowest oit (Almaz 2006. Vessels avigatig through the Strait have to make may shar turs (betwee 45 a eve 80 egrees which carry high risks for the vessels i such a arrow chael (Ulusçu et Al. 2009. The Strait which is situate i the mile of a huge metroolita area of 15 millio resiets, features a very heavy maritime traffic (more tha 51,000 vessels aually, with more tha 15,000 such vessels carryig agerous cargo; there is also heavy local traffic icluig more tha 2,000 asseger ferry tris aily betwee the two shores (Göültaş 2007. Oe oteworthy roerty of the Strait is the revailig currets which may rise u to 8 kots see. Other averse meteorological coitios like fog, wi, rai a storm also icrease the ifficulty of avigatio i the Strait. I ese fog coitios, vessel traffic may be artially or wholly susee util meteorological coitios imrove, which causes agerous a uwate ile-us at the Strait etraces a uts further strais o the maritime traffic maagemet, sice it icreases avigatio roblems (Özbaş 2005. The Vessel Traffic System (VTS was establishe i 2004 i orer to regulate a guie maritime traffic i the Strait, i accorace with iteratioal a atioal covetios a regulatios, while imrovig safe avigatio, rotectig life a eviromet. Withi the framework of this system, vessels esirig to trasit the Strait have to submit two reorts to the VTS, Sailig Pla 1 (SP-1 a Sailig Pla 2 (SP-2. SP-1 iclues all the iformatio about the vessel a must be submitte at least 24 hours before the arrival. SP-2 is of vital imortace for laig of vessel assages from the Strait a must be submitte at least 2 hours or 20 autical miles (whichever comes first rior to etry ito the Strait. The VTS aalyze the ata i these reorts a reare a safe aily sailig traffic la (VTS Users Guie. 2. SIMULATIO MODEL The first ste to better uersta the risks geerate by the maritime traffic i the Strait is to uersta a moel the maritime actively i the Strait. This stuy aims to esig a evelo a simulatio moel to rereset the actual traffic flow i the Strait with regar to the VTS rules a regulatios (R&R a olicies that meteorological a geograhical coitios, suort services (like ilot a tugboats a frequecy, tye a cargo characteristics of vessel arrivals (to make a assage through the Strait with the aim of ietifyig the imact of such factors o traffic coitios, otetial roblems a bottleecks for a less risky trasit a overtakig allowace urig the assage of vessels o Strait laes. 2.1. Vessel Classificatio The VTS has a secific vessel classificatio system base o vessel tyes, cargo characteristics a vessel legths. I this stuy a somewhat simlifie versio of this classificatio (which is islaye i Figure 1 is use. The mai reaso why takers a agerous cargo vessels u to 100 meters a LPG-LG u to 150 meters, takers a agerous cargo vessels betwee 100 a 150 meters a ry cargo carryig vessels betwee 150 a 300 meters are lace i the same class is that accorig to the VTS regulatios, they have to satisfy the same coitios i eterig a avigatig the Strait. This way of classificatio simlifies the uerstaig of vessel etrace a sailig coitios. 439
Figure 1: Vessel Classificatio 2.2. The Arrival Process The Area Iut Aalyzer which is a very efficiet tool for istributio fittig to ata is eloye i fittig iterarrival time istributios. Via the Iut Aalyzer s Fit meu, all robable istributios fitte to the actual ata are reveale a fit all roerty estimates the istributio with the miimum square error. After fittig a istributio, a histogram a the robability esity fuctio (f suerimose o the histogram summarize the characteristics of the fit (Law a Kelto 2007. To illustrate, the best fitte iterarrival time istributio of orthbou Class E vessels is fou as the Gamma istributio with shae arameter α beig 648 a scale arameter β beig 0.974. I the summary reort of Area Iut Aalyzer (as islaye i Figure 2, the shae of the robability esity fuctio overlas with the histogram a just lookig at this figure, oe gets the feelig that the selecte fuctio reresets the actual iterarrival time ata quite well. Figure 1: Histogram of orthbou Class E iterarrivals 2.3. The Istabul Strait Traffic Rules a Regulatios Vessels eter the Strait either from the orth, (travelig south a thus are calle as southbou vessels or from the south (travelig orth a thus are calle orthbou vessels etraces. Some R&R relate to vessel trasit maagemet that are also reflecte i the simulatio moel are as follows: There shoul be at least a 10-miute iterval betwee two cosecutive reay to eter vessels from oe irectio. Class A a T6 vessels ass through the Strait oly urig aytime. o vessels are allowe to meet with Class A vessels. Class, C a E vessels shoul ot meet each other urig bi-irectioal ighttime flow. There shoul be at least 75 miutes betwee two cosecutive southbou Class A vessels a at least 90 miutes betwee two cosecutive orthbou Class A vessels. Passeger vessels are allowe to the Strait regarless of their irectio of flow whe ursuit istace, meteorological a ilot a tugboat request coitios are satisfie. Southbou stoover vessels have riority over orthbou stoover vessels, which have riority over ay o-stoover vessels. 2.4. Vessel Sequecig Observatios of the 2009 trasit ata a iscussios with the VTS authorities have iicate that the imlemetatio of the regulatios regarig ursuit istaces betwee two cosecutive vessels of various classes ca be arameterize ito a set of easily followe rules. Let θ be the miimum ursuit istace betwee two cosecutive vessels of class D, E, P travelig orthbou a let µ be the miimum ursuit istace betwee two cosecutive vessels of class D, E, P travelig southbou. Accorig to the R&R, the miimum ursuit istace betwee a orthbou (southbou class D, E or P vessel a a class A, or C vessel sailig i the same irectio is also θ (µ. The miimum ursuit istace betwee two cosecutive class C vessels travelig orthbou (southbou is 2*θ (2*µ a the miimum ursuit istace betwee a orthbou (southbou class C a a class A or vessel sailig i the same irectio is also 2*θ (2*µ. The miimum ursuit istace betwee two cosecutive A a vessels travelig orthbou (southbou is resectively 6*θ (6* µ a 4*θ (4* µ. 2.5. Daytime Vessel Scheulig As metioe before, traffic flows from oe irectio at a time urig aytime. The maximum uratio of aytime a start time of the aytime traffic iffer accorig to seasos. The first irectio of vessel flow 440
ito the Strait at aytime is etermie base o the total umber of vessels i queues a their waitig time regarig vessel riorities (two hours before the startig time. The formula use for i the etermiatio of startig irectio is as follows: Oosite irectio traffic time wiow legth is calculate as: W (3 S (1 where: S : score value of the active irectio S : score value i the oosite (assive irectio α: multilicative costat for umber of vessels i queues b : multilicative costat for waitig time of vessels i queues Ca : coefficiet for A tye vessels Cc : coefficiet for C tye vessels C : coefficiet for D tye vessels Ce : coefficiet for E tye vessels ( : umber of i tye vessels i queue i active Q ( i t s irectio at time t=t s ( : umber of i tye vessels i queue i Q ( i t s assive irectio at time t=t s ( WT ( j : total waitig time of j tye vessels i active t s irectio at time t=t s ( : total waitig time of j tye vessels i assive WT ( j Ca * a * Ca * WT ( b* WT ( t s ( ( ( ( Cc * C C WT ( C irectio at time t=t s ( ( Cc * WT ( C ( ' ( This formula is alie for both irectios a the irectio with higher score is eclare as the startig irectio of the aytime traffic scheule. Two sigificat factors ifluecig the etermiatio of the first irectio of aytime flow are the umber of vessels i queues a vessel waitig times a they are i ifferet level of sigificace. (The associate weights α a are omiate as 0.25 a 0.75 resectively. Class A a T6 vessels are the most critical vessels i terms of the risks they geerate. Therefore, i orer to set out the framework for aytime scheule, (after attaiig the first irectio of aytime traffic, umber of Class A vessels trasitig from both irectios are estimate. I this resect, maximum aytime uratio is ivie ito two, roortio to the umber of Class A vessels i orthbou a southbou queues. Startig irectio traffic time wiow legth is calculate as: (2 W C * D ( D C * WT ( D WT ( D ( ( ( Ce* E ( E Ce* WT ( E WT ( E ( ( ( The umber of Class A vessels lae to eter the Strait urig the startig irectio vessel traffic flow is: W ( (4 6 * ( where: if is orthbou ( (5 if is southbou where: The arameters i the eomiator chages with regar to startig irectio ecisio. The umber of Class A vessels lae to eter the Strait urig the oosite irectio vessel traffic flow is: W (6 ( ' 6 * ( ' oth ( a (' to earest iteger umbers. are roue ow Waitig time of vessels is ajuste eeig o whether they are stoover vessels or ot. The ajuste waitig time of vessel j is efie by: c 1.5 1.25 1 W if if a j c* WT ( j ts j is a stoover southbou j is a stoover orthbou vessel otherwise vessel (7 (8 Sice asseger vessels have the highest riority i vessel sequecig, the moel first searches the Class P queue i the etermie irectio. If there exist ay P vessels i the etermie irectio a if the visibility coitios a ilot a tugboat ema are satisfie, the oe havig the maximum elase waitig time is allowe to the Strait a the time is icremete as θ (µ miutes. Meawhile, if there exist ay P vessels o the other sie, the oe with the maximum elase waitig time is allowe to the Strait as well (eve though a ui-irectioal time wiow is i actio. If there is o P vessel i the etermie irectio, the moel searches the Class A queue. If there is ay A tye vessel i the etermie irectio, the the ursuit istace requiremets, meteorological situatios a ilot a tugboat availabilities are checke. Whe all coitios are fulfille, the class A vessel havig the 441
maximum elase waitig time eters the Strait, otherwise moel examies the Class C, E a D vessel queues resectively a allows the oe havig maximum elase waitig time regarig their miimum ursuit istaces amog class tyes. As soo as a vessel eters the Strait, agai time is icremete as the miimum ursuit istace iterval (as θ or µ miutes a the other istace rules amog vessel tyes are also checke util the last lae A vessel i the active irectio eters the Strait. Sice the origial aily scheule is mae i the morig (two hours before traffic start time, the ui-irectioal time wiows of that scheule are esigate to service just the available vessels (esecially A vessels at that time. So, close to the e of the time wiow of the startig irectio, say at time t= t, the moel reviews the umber of Class A vessels i queues a revises the origial scheule to exte the ui-irectioal time wiows as log as the maximum aytime uratio ermits. This extee time iterval is ame as the slack time. For slack time traffic la, the umber of Class A vessels lae to eter the Strait urig the startig a oosite irectio ui-irectioal traffic flow time wiows is comute by iviig this aortioe times by the miimum ursuit istace betwee two cosecutive Class A vessel trasitig from startig a oosite irectios time wiows. The legth of slack time is: ST MAX( 0, DT ( t t s W (8 where ts is the start time of the first irectio vessel traffic flow. The stes for slack time scheule at time t= t are as follows: (i umber of Class A vessels i the oosite irectio at time t= t is checke. Oe imortat etail at this oit is igorig the umber of reviously lae vessels i the oosite irectio ( (, sice they are alreay scheule to ass i the origial time wiow etermie at la time. amely, the ew arrivals (sice la time of class A vessels i oosite irectio are: Q ( t MAX(0,( t ( ' (9 (ii The aitioal waitig time of ew arrival (sice la time class A vessels i irectio at time t is comute. This ca be oe by removig the realize waitig time of lae A vessels from total waitig time of Class A i irectio, that is: ' ( ' ' WT ( WT ( WT ( (10 ts (iii The ratio for umber of uscheule class A vessels i both irectios is estimate as: ( X (11 Sice t reresets a time oit at which all scheule vessels i the active irectio have alreay move ito the Strait, the umerator must oly cotai the ew arrival class A vessels sice la time. i The ratio for waitig time of uscheule vessels i irectio a at time t is ( WT ( Y WT ( calculate as: (12 ii If the amout of slack time is larger tha or equal to time legth that allows a southbou A vessel trasit (6*, the slack time algorithm tries to make use of this time by scheulig oe more orthbou or southbou class A vessel. iii The iicator Z is etermie as follows: Z X * Y * b (13 iv The exact roceure of allocatig the slack time to aitioal orthbou a / or southbou class A vessels is as follows: a. If Z is greater tha or equal to 1, it is euce that the aitioal class A vessel (lae to ass i the slack time shoul be a -irectioal vessel a the the equatios (11 a (12 are uate. umber of -irectioal lae A vessels i slack time ( ( is icremete by oe. ( ( 1 (14 a the slack time legth is uate as: ST ST 6* ( (15 b. If Z is less tha 1, it is euce that the aitioal class A vessel (lae to ass i the slack time shoul be a -irectioal vessel a the the equatios (13 a (14 are uate. umber of -irectioal lae A vessels i slack time ( is ( ' icremete by oe a the slack time legth is uate same as equatio (15. 442
( ' ( ' 1 (16 (viii Returig to ste (iii, the algorithm rocees util the e of ST. y meas of this rescheule roceure, more vessels from both irectios are scheule a amitte to trasit util the e of the slack time. At the e of the (extee startig irectio time wiow (i.e. with the etrace to the Strait of the last scheule class A vessel from that irectio, the traffic is close from both irectios util the last vessel leaves the Strait. Sice it takes aroximately 30 miutes for a class A at Filburu (i orthbou traffic flow case or at oğaziçi rige (i southbou traffic case to comletely exit the Strait, the time ga betwee the last orthbou or southbou Class A vessel a the followig vessel from the oosite irectio shoul be 6* θ+ 30 or 6* µ+30 miutes, resectively. At the e of the startig irectio time wiow (i.e. with the etrace to the Strait of the last scheule class A vessel from that irectio, the traffic is close from both irectios util the last vessel leaves the Strait. The start a executio of the vessel traffic flow i the oosite irectio traffic is the same as the first irectio flow. Vessels are allowe ito the Strait util reachig the umber of lae A vessels i this irectio. If slack time amits ay more A vessels i this irectio, they also eter the Strait util the start of the ighttime vessel traffic. A tyical examle for aytime vessel scheule is islaye i Figure 3. orthbou Southbou eciig wiows legth after sequecig class vessels at ighttime la (t= t are uate i startig a oosite irectios resectively as follows: Qt ( u Qt ( MAX(0,( Qt ( C ( Qt ( 1/ 2 (17 ' ' ' Qt ( u Qt ( MAX(0,( Qt ( C ( Qt ( 1/ 2 The, the tetative time wiow legth i the ighttime active irectio is calculate as follows: ( Qt ( u W T * (18 Q ( Q ' ( t The tetative time wiow legth i the ighttime assive irectio is calculate as follows: Qt ( u (19 u W ( ' T * Q ( Q ' ( where T is the total ighttime uratio, which is the time ga betwee the followig ay s aytime traffic la start time a the e of the reset ay s aytime wiows. Accorigly, the umber of Class vessels lae to eter the Strait i the active irectio flow is: ( mi( Q t t u t W (, 4* u t ( ( u (20 The umber of Class vessels lae to eter the Strait i the assive irectio flow is: A D D C D D A E D C D E A Daytime Start Slack time flow Start ( ' mi( Q t W (, 4 * ( ' ( ' (21 Southbou orthbou A P C D D D A D D C D E A Figure 3: Daytime Scheule 2.6. ighttime Vessel Scheulig Whe aytime traffic es, the active traffic flow irectio remais as the first (active irectio of ighttime traffic. Aitioally, ulike aytime uiirectioal traffic, at ighttime, there exist two restricte vessel flows (accorig to the R&R, Class D vessels may eter from the oosite irectio whe there are such vessels available a meteorological coitios allow sice o Class A vessels are allowe from either irectio urig ighttime. umber of Class vessels a the umber of all Class C vessels (the oes which will be use for Presumig Class the most critical grou i the ighttime scheule, the legth of the orthbou a the southbou time wiows are outlie by Class vessels (similar to the role of Class A vessels i aytime scheulig. However, the relatively high oulatio of the abuace of Class C vessels (arou 9000 Class C vessels i a year ecessitates the cosieratio of this class while esigig the ighttime traffic la. Cosierig that miimum ursuit istace betwee two Class vessels is 4*θ (4*µ whereas miimum ursuit istace betwee a Class vessel a a Class C vessel is 2*θ (2*µ, the uratio of ighttime restricte traffic flow time is etermie by the umber of lae Class vessels (multilie by 2*θ or 2*µ, accorig to the active irectio, a the umber of remaiig class C vessels (multilie by θ or µ, accorig to the active irectio. 443
The total umber of Class C vessels lae to eter the Strait after sequecig class vessels i the active irectio flow is: Qt ( C ( ( 1 max( 0, 2 (22 ( C ( The total umber of Class C vessels lae to eter the Strait after sequecig class vessels i the assive irectio flow is: Qt ( C ( ( ' 1 max( 0, 2 (23 ( C ( ' oth equatios (22 a (23 are roue ow to earest iteger umbers. The resultig total ighttime vessel traffic uratio i the active irectio is: C W( mi( W (, ( *4* ( ( *4* ( (24 The resultig total ighttime southbou vessel traffic uratio i the assive irectio is: C W( ' mi( W ( ', ( '*4* ( ' ( '*4* ( ' (25 Oce the scheule trasit of Class a C vessels is comlete, if there is remaiig ighttime, Class D a E vessels cotiue eterig the Strait from both irectios (with Class E still havig higher riority accorig to the miimum ursuit istaces (θ or µ rules. 2.7. The Traffic Laes a Overtakig I the moel, vessels follow two mai laes, (the orthbou or the southbou laes a the overtakig lae, if ermitte, while trasitig the Strait. The whole Strait is ivie ito 22 slices with statios. Slices are at eight cables (0.8 autical miles 1.482 km. itervals a i orer to sustai a reetermie ursuit istace betwee vessels each slice is also comose of 2 cables log substatios. Sice stoig i the Strait for ay reaso is ot allowe, vessels cotiuously move from oe statio to aother urig their stay i the Strait. Overtakig is allowe i the Strait excet at the arrowest art, accorig to these coitios: Whe a vessel is i the overtakig lae, there shoul be o other vessel i this lae i the oosite irectio at least u to the ext statio. There shoul be at least the ursuit istace betwee two closest vessel i the overtakig lae travelig the same irectio. After overtakig is comlete, vessels move back to the mai laes. 2.8. Pilot a Tugboaervices Accorig to the R&R, havig a ilot catai o boar urig the Strait assage is comulsory for vessels loger tha 250 meters a otioal (though strogly recommee for other vessels. All vessels exress their ilot catai a tugboat ees i their SP-1 a SP-2 reorts. There are 20 ilots a 6 tugboats available(as i the real situatio. I the simulatio moel ilots a tugboats are treate as resources which are seize by vessels at the embarkig area i the Strait a release while leavig. I orer to meet ilot a tugboat ees, every hour the moel searches the umber of available ilots (icluig trasferrig ilots i the active irectio a requests ilots from the oosite sie whe it is less tha 6. The moel also searches the umber of available tugboats i the active irectio a requests tugboats from the oosite sie whe it is less tha 3. Durig the ighttime time wiows, umber of ilots at both sies is equalize to 6 a tugboats to 3 to meet the ilot a tugboat ema. Oce a ilote vessel s assage i a certai irectio is comlete, the ilot is release from its curret uty a iclue i the set of available resources for the oosite irectio. 2.9. Visibility Coitios Accorig to the R&R, whe visibility is less tha oe autical mile i the Strait (calle FogTye1, oly oeway traffic is ermitte a whe visibility i the Strait is less tha 0.5 mile (calle as FogTye2, vessel traffic is susee i both irectios. The visibility moule i the simulatio moel reas the fog iformatio from the visibility ata of (Almaz 2006 exterally. efore a vessel is allowe to eter the Strait from the active irectio urig aytime, visibility coitio is checke; if there is a FogTye2 evet, the vessel waits util it isaears. FogTye1 oes ot affect aytime flows very much (sice almost all vessel activity with the excetio of class P vessels is ui-irectioal ayway; oly the class P vessels comig from the oosite (assive irectio are stoe. Whe a FogTye1 occurs at ighttime, however, two-way traffic is susee. 2.10. Curret Coitios The most omiat curret tye o the Strait is the southbou surface curret cause by level ifferece betwee the lack Sea a the Meiterraea Sea. The curret moule of the simulatio moel is itegrate ito the moel from the revious stuy (Almaz 2006. I the stuy, the most effective southbou curret is take ito accout a a movig average fuctio is built to estimate a aily base curret value. The, the curret level at ifferet regios of the Strait are assige as reetermie ercetages of the base value, base o historical curret ata. I orer to comly with the R&R, whe curret see excees 4 kots, class A,, C a E vessels havig a see less tha 10 kots are ot allowe ito the Strait. Moreover, all vessels i these classes have to wait i their queues 444
(util curret coitios stabilize whe curret see excees 6 kots. 3. OUTPUTS OF THE MODEL This moel is ru for the 13 moths time erio (betwee 1 December 2008 a 1 Jauary 2010. The first moth is cosiere as the warm u erio. Some erformace measures etermie for the aalysis are: R1:The average waitig time of vessels (aggregate a vessel tye base; R2:Total umber of vessels asse; R3:Average umber of vessels i queues; R4:The etire Strait vessel esity; R5: Pilot utilizatio; 4. VERIFICATIO AD VALIDATIO Due to the fact that the simulatio moel i this stuy cosists of may submoels itegrate ito the mai traffic moel ruig cocurretly, it is ifficult to moitor the system. However; with the trace moule of Area, arrival of each vessel, attributes assige to it, its movemet to the achorage area or to the aroriate queues a its amittace to the Strait are followe clearly, while simultaeously watchig etities relate to meteorological evets affectig the system. Moreover, aimatio reveals all evets i the whole system; therefore, logic errors ca be cature easily. Variable iicator of the Area is also a frequetly utilize tool i this stuy. The chage i values of erformace measures ca irectly be trace by variable iicators. Extreme coitio verificatio is first erforme by icreasig vessels arrival rates by 20% i a three moth simulatio ru. Whe comare to the base sceario, average vessel waitig time shows more tha fifteefol icrease (from 541 miutes to 9272 miutes, average umber of vessels i queues icrease from 52.6 to 1154.4, umber of vessels asse icreases to 14756 from 12845 a ilot utilizatio icreases from 0.23 to 0.25. Aother extreme coitios effect is reucig the total umber of ilots i the moel to 12 istea of 20. The moel is ru for oe year with 25 relicatios a as execte, the ilot utilizatio, average, maximum waitig time of vessels a umber of vessels i queues icrease a total umber of vessels asse the Strait ecrease. The most coclusive of the valiatio tests i this stuy are the outut comarisos with the real 2009 ata. The results of selecte erformace measures are sufficietly close to the ata 2009 to suort the claim that the moel mimics the actual system reasoably well. As a examle, average waitig times of all vessels i moel a i actual ata are comare. The results are quite similar to each other, as islaye i Table 2. Table 2: Comariso of average waitig time of vessels Waitig Times (i miutes The Simulatio Moel 2009 Data Half Relative Average With Error (% All Vessels 842 814.4 123.13-3.28 5. SCEARIO AALYSIS AD RESULTS Four factors are selecte for the sceario aalysis of the simulatio moel: A: miimum ursuit istace (i time uits betwee vessels : vessel rofile C: ilot olicy D: arrival rate The levels of ietifie factors for sceario aalysis are islaye i Table 3. Table 3: Mai factors a their levels i sceario esig Factor ame Low Average High A ursuit istace 13-11.5S 13.5-12S 14-12.5S vessel rofile base base >=150 m C Pilot availability 16 20 24 D arrival rate base 5% more 10% more The first factor A with three levels is the miimum ursuit iterval betwee two cosecutive vessels (13 for the low settig meas 13 miutes iterval for orthbou vessels a 11.5S meas 11.5 miutes iterval for southbou vessels. Regarig the vessel rofile factor, the low settig corresos to the base sceario i which vessels ema ilots accorig to the ilot request frequecy istributio of vessel classes geerate base o the 2009 ata. I the high settig, i aitio to this raom ilot ema, all vessels loger tha 150 meters are routiely assige a ilot while assig the Strait. I ilot availability factor C, the umber of available ilots is set at 16 for the low level a 20 for the average level (as is the case i the curret system a 24 for the highest level. Accorig to the last factor, regarig the arrival rate of vessels D, the low settig (which is the settig assume i the base sceario is take as the rates estimate i the iterarrival istributio for each subclass base o the 2009 ata. I the average level, arrival rate of vessels is icrease by 5 er cet (comare to the rates estimate base o the 2009 ata a i the high level, vessel arrival rates are icrease by 10 er cet. Accorigly, a total of 54 ifferet scearios (icluig the base sceario, are rojecte a ru with 25 relicatios for a full factorial esig. The oututs of these scearios are gathere from Area reorts, the sigificat factors a their iteractios are ivestigate through the AOVA tables i the Desig Exert 8.0 software. The ercet cotributio of each factor o erformace measures are islaye i Table 4. 445
Table 4: Percet cotributios of mai factors A C D AD Average waitig time 38 59 2.4 Total vessels asse 0.3 0.1 0.1 98 0.1 Average trasit time 89 Pilot utilizatio 0.3 3.1 93 3.5 Vessel esity 0.3 0.1 0.1 99 0.1 I orer to track the effects of factors easily, sigle factor level chage i scearios is ivestigate through the comariso of scearios 19, 3, 7 a 16 with the base sceario 1 as ca be see i Table 5. Table 5. Scearios with various factor level chages Scearios R1 R2 R3 R4 R5 1 814 51,178 79.9 9.45 0.24 19 608 51,206 59.2 9.46 0.24 3 722 51,204 70.3 9.46 0.31 7 754 51,200 73.4 9.45 0.25 4 2289 56,628 251 10.5 0.26 10 2275 56,624 250 10.5 0.27 12 2170 56,677 237 10.5 0.22 25 663 51,193 64.6 9.45 0.25 49 614 53,880 62.8 9.95 0.22 52 622 53,882 63.7 9.95 0.25 Decreasig ursuit istace to 13.5 miutes for south etraces a to 12 miutes for orth etraces (sceario 19, rimarily ecrease the waitig time (by 25 er cet, ecrease the umber of vessels i queues by 26.25 er cet, while keeig the total umber of vessels asse a vessel esity almost the same. Decreasig the umber of available ilots from 20 to 16 (sceario 3 icreases ilot utilizatio by 29.2 er cet a ecreases waitig time by 11.30 er cet (the reaso why the average waitig time ecreases is ue to ecrease i waitig time of Class D vessels, which eter the Strait more frequetly while other vessel tyes remai waitig because of ilot uavailability. Assigig ilots for all vessels loger tha 150 meters (sceario 7 icreases ilot utilizatio by 4.1 er cet. Icreasig vessel arrival rate by te er cet (sceario 4 icreases total umber of vessels asse by 10.64 er cet, average waitig time by 181 er cet, umber of vessels i queues by 212 er cet, ilot utilizatio by 29.2 er cet a vessel esity by 10.8 er cet. The effect of two, three a four factor level chages over resoses is also ivestigate. For istace, ecreasig ursuit istace to 13.5 miutes for orthbou a to 12 miutes for southbou vessels while assigig ilot for all vessels loger tha 150 meters (sceario 25 ecrease the waitig time by 18.6 er cet whe comare to the base sceario; however, waitig time is icrease by 9.9 er cet whe comare to the sigle factor level chage case ivolvig 13.5 miutes ursuit istace for south etraces a 12 miutes for orth etraces (sceario 19. Table 5 also islays that icreasig vessel arrival rate by te er cet a 20 available ilots to 24 i the system while assigig ilot for all vessels loger tha 150 meters (comariso of scearios 10 a 12 ecrease the waitig time by 4.62 er cet a umber of vessels i queues by 5.22 er cet. Furthermore, icreasig 20 available ilots to 24 i the system while assigig ilot for all vessels loger tha 150 meters uer five er cet higher arrival rate (comariso of scearios 49 a 52 have almost same erformace measure results. I aother sceario aalysis, 4 factors ifluecig the resose variables uer high arrival rate coitios (umber of arrive vessels icrease by 10 er cet are aalyze. The levels of factor A are 13.5 miutes for orthbou a 12 miutes for southbou i low settig a 14 miutes for orthbou a 12.5 miutes for southbou i high settig. Regarig the visibility factor (D, the low settig escribes the base sceario i which vessels ecouter fog evets accorig to the visibility submoel, whereas i the high settig, the fog atter of the worst case (i.e. the autum fog realizatios which have the logest fog uratios is chose as the visibility ata for the whole year. I orer to track the effects of factors easily as islaye i Table 6, level chage i scearios is ivestigate comare to the base sceario 1. Decreasig ursuit istace to 13.5 miutes for orth etraces a to 12 miutes for south etraces (sceario 7 rimarily ecrease the waitig time by 41.5 er cet, ecrease the umber of vessels i queues by 41.6 er cet, while keeig the total umber of vessels asse almost the same. Settig low visibility coitios (sceario 13 icreases average waitig time by 88.8 er cet yet oes ot sigificatly chage the total umber of vessels asse. The effect of two a three factor level chages over resoses may also be ivestigate i this table. For examle, although reucig ursuit istaces to 13.5 miutes for orthbou assages a 12 miutes for southbou assages a eloyig 24 ilots istea of 20, the average waitig time icreases by 82 er cet uer low visibility coitios (comariso of scearios 7 a 20 a umber of vessels i queues icrease by 91 er cet. Table 6. Scearios with various factor level chages uer high arrival rate coitios Scearios R1 R2 R3 R4 R5 1 2289 56629 250 10.5 0.3 7 1367 56850 146 10.5 0.3 13 4320 56324 474 10.4 0.3 15 4087 56336 448 10.4 0.4 20 2482 56531 279 10.5 0.2 23 2354 56615 262 10.5 0.2 I the full factorial aalysis of the relate scearios, the 24 ifferet scearios are exerimete through 25 relicatios (i.e. the sceario aalysis is comose of 600 istict observatios. I the sceario aalysis, the 446
most effective factor o erformace measures is observe as visibility coitios. As fog i the Strait becomes stroger, average waitig time of vessels a trasit time icrease. Moreover, low visibility coitios ecrease total umber of vessels asse from irectios, ilot utilizatio a vessel esity i the Strait. 6. COCLUSIO AD FURTHER RESEARCH I this stuy, a simulatio moel is eveloe for reresetig the vessel traffic behavior i the Strait. I this simulatio moel, maritime rules a regulatios about vessel amittace, ursuit istaces amog vessels, riority levels of istict vessel tyes a ilot requiremets are all cosiere. Moreover, submoels reresetig meteorological coitios such as fog, curret a storm are itegrate to the moel. For valiatio uroses, the simulatio oututs are comare with the actual 2009 ata a quite satisfactory results are obtaie. I orer to aalyze the effects of various factors such as vessel arrival rate, vessel rofile, ilot availability a miimum ursuit istaces betwee vessels, o erformace measures, 54 scearios are erforme with the full factorial esig. The most sigificat factor for all selecte variables is observe as the vessel arrival rate. The miimum ursuit istace betwee vessels is also sigificat for most erformace measures. The iteractio of arrival rate a ursuit istace is effective o the most resoses, as well. Pilot availability is ricially imortat for ilot utilizatio. Aother sceario aalysis is coucte whe vessel arrival rate is icrease by 10 er cet a the visibility factor is ae. Results associate with the cosiere 24 scearios show that visibility is the most critical factor for erformace measures a its iteractio with miimum ursuit istace at ifferet levels is also sigificat for erformace measures such as average waitig time of vessels, umber of vessels asse a ilot utilizatio. This stuy is lae to be use for risk aalysis of the Strait. Icororatig robable vessel acciets a the cosequeces to the moel ca have a very beeficial effect for revisig the olicies a miimizig risk. Özbaş,. 2005, Simulatio of Maritime Trasit Traffic i the Istabul Chael, M.S. Thesis, Deartmet of Iustrial Egieerig, oğaziçi Uiversity, Istabul. Ulusçu, S. Ö.,. Özbaş, T. Altıok a İ. Or. 2009. Risk Aalysis of the Vessel Traffic i the Strait of Istabul, Risk Aalysis, Vol. 20, o. 10,. 1454-1472. VTS Users Guie, Turkish Straits Vessel Traffic Service. 2004. Geeral Maagemet of Coastal Safety a Salvage Amiistratios, 3r eitio, Istabul. AUTHOR IOGRAPHIES İLHA OR was bor i Istabul, 1951. He receive his S (1973, MS (1974 a Ph.D. egrees (1976 from orthwester Uiversity, Evasto, Illiois, USA. He has bee a faculty member at Deartmet of Iustrial Egieerig of ogazici Uiversity, Istabul, Turkey sice 1976. He was a visitig faculty member at Syracuse Uiversity (1982-1983 a Uiversity of Maryla (1983. He serve o the aval Research Logistics Quarterly Joural s Eitorial oar betwee 1993 a 2003.Research areas are evirometal a risk maagemet, eergy olicy a laig, rouctio a maiteace laig. His e-mail aress is: or@bou.eu.tr a ersoal web age is: htt://www.ie.bou.eu.tr/~or/ ŞİRİ ÖZLEM was bor i ursa, 1985. She receive her S (2008 from Uluag Uiversity, ursa,turkey a MS (2011 from ogazici Uiversity, Istabul, Turkey. She is curretly a octoral stuet at the Deartmet of Iustrial Egieerig of oğaziçi Uiversity, Turkey. Her e-mail aress is: siri.ozlem@bou.eu.tr İRUR ÖZAŞ was bor i Istabul, 1980. She receive her S (2003 i Systems Egieerig from Yeitee Uiversity, Istabul, Turkey MS (2005 a PhD (2010 i Iustrial Egieerig from ogazici Uiversity, Istabul, Turkey. She is curretly a ost octoral associate at Ceter for Avace Ifrastructure a Trasortatio (CAIT / Laboratory for Port Security (LPS, Rutgers, The State Uiversity of ew Jersey,U.S.A. Her e-mail aress is: birur@ozbas.com.tr REFERECES Almaz, A. Ö., 2006, Ivestigatio of the Maritime Trasit Traffic i the Istabul Chael through Simulatio Moelig a Sceario Aalysis, M.S. Thesis, Deartmet of Iustrial Egieerig, oğaziçi Uiversity, Istabul. Göültaş, E., Aalysis of the Extreme Weather Coitios, Vessel Arrival Processes a Prioritizatio i the Strait of Istabul through Simulatio Moelig, M.S. Thesis, Deartmet of Iustrial Egieerig, oğaziçi Uiversity, Istabul, 2007. Law, A. M. a W. D. Kelto. 2007 Simulatio Moelig a Aalysis, McGraw-Hill Press, Sigaore. 447