ABSTRACT
Title of thesis: TRANSIENT FIRE LOAD
ON ALUMINUM FERRIES
Brian Hall, Master of Science, 2014
Thesis directed by: Professor J. Michael Gollner
Department of Fire Protection Engineering
The transient fire load aboard aluminum passenger ferries is studied to de-
termine the contribution that baggage has on increasing the temperature of the
compartment overhead, which serves as the deck for passenger rendevous during
fire emergencies on many large vessels. Single-point and average temperature max-
imums are compared for a variety of baggage fire scenarios to determine if critical
temperatures are reached that would compromise the structural integrity of the
aluminum.
A survey of passenger ferry vessels has been performed to determine the extent
and type of baggage loading present in passenger compartments. The baggage type,
carriage rate, and baggage weight were recorded to determine the overall fire load as
well as the average weight of luggage brought on board. Ferry vessels were examined
for problem locations and potential sources of elevated flame lengths that may cause
the flame to impinge directly on the aluminum structure overhead.
The Fire Dynamics Simulator (FDS) by the National institute of Standards
and Technology (NIST) is used to model a representative large passenger ferry com-
partment. Multiple scenarios are simulated with baggage and seat burning along
with consideration of flame spread based on a critical heat flux and collected survey
results.
Based on the results of the survey, it was determined that the majority of
aluminum ferries, when fully loaded, attain higher fuel loads than allowed by current
Coast Guard requirements. Subsequent simulations also revealed that the current
level of loading compromises the structural integrity of the aluminum superstructure
on an average ferry. Additional scenarios tested, such as a stroller parked in the
corner of a passenger compartment, would raise the temperature of the aluminum
superstructure to a level that would compromise safety. It is recommended that
regulatory changes be made to ensure that these severe scenarios are avoided to
protect life and property.
TRANSIENT FIRE LOAD ON ALUMINUM FERRIES
by
Brian Hall
Thesis submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial fulfillment
of the requirements for the degree of
Master of Science
2014
Advisory Committee:
Professor Michael J. Gollner, Committee Chair
Professor James Milke
Professor Stanislav Stoliarov
c? Copyright by
Brian Hall
2014
Acknowledgments
I would like to thank my wife and family for their support and for being so
flexible for the last two years with my constantly shifting schedule.
I would like to thank my adviser Dr. Michael J. Gollner for his guidance and
support throughout this research endeavor. He has been instrumental in ensuring
that I am on track with my research.
I would also like to extend my gratitude to Rick Nolan of Boston Harbor
Cruises, Christian Myers and William McCombe of Interstate Navigation, Greg
Gifford of Interstate Navigation, and the captains and crews of the surveyed vessels.
Everyone went out of their way to ensure that I had full access to the vessels and
was able to collect all of the information needed for this study. In addition, John
Duclos, of Gladding-Hearn Shipbuilding, was instrumental in providing me with
builder information on aluminum vessels to ensure accuracy of my research.
Finally I would like to thank Carl Moberg, Tom Woodford, LT Noel Shriner,
LCDR John Miller, and all the other Coast Guard personnel who assisted me and
provided me with insight into this research topic.
ii
Table of Contents
List of Tables v
List of Figures vi
List of Abbreviations ix
1 Introduction 1
1.1 Regulatory History . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Past Fires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4.1 Scandanavian Star . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4.2 Al Salam Boccaccio 98 . . . . . . . . . . . . . . . . . . . . . . 7
1.4.3 Lady Martha . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Literature Review 11
2.1 Vessel Inspection History . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.1 Aluminum Vessels . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.2 Coast Guard Research . . . . . . . . . . . . . . . . . . . . . . 17
2.1.3 Regulatory Changes . . . . . . . . . . . . . . . . . . . . . . . 22
2.2 Recent Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2.1 Coast Guard FDS Modeling . . . . . . . . . . . . . . . . . . . 25
2.2.2 Swedish Baggage Testing . . . . . . . . . . . . . . . . . . . . . 31
3 Carriage Rates 36
3.1 Allowed Carriage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2.1 Commuter Ferry . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.2.2 Non-Commuter Ferry . . . . . . . . . . . . . . . . . . . . . . . 41
3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
iii
4 FDS Model 48
4.1 Verifying the Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2 Grid Resolution Study . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.3 Creating an Average Bag . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4 Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.4.1 Stroller in the Corner . . . . . . . . . . . . . . . . . . . . . . . 58
4.4.2 Burning of Two Seat Banks . . . . . . . . . . . . . . . . . . . 59
4.4.3 Burning of a Single Seat Bank . . . . . . . . . . . . . . . . . . 62
5 Results & Analysis 64
5.1 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.1.1 Stroller in the Corner . . . . . . . . . . . . . . . . . . . . . . . 64
5.1.2 Burning Bag In Aisle . . . . . . . . . . . . . . . . . . . . . . . 67
5.1.3 Single Burning Seat Bank . . . . . . . . . . . . . . . . . . . . 71
5.2 Analysis of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.2.1 Evacuation Carriage . . . . . . . . . . . . . . . . . . . . . . . 74
5.2.2 Extent of Simulations . . . . . . . . . . . . . . . . . . . . . . . 76
5.2.3 Practical Considerations . . . . . . . . . . . . . . . . . . . . . 77
6 Conclusions and Recommendations 80
6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.2 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Appendices 83
A Vessel Survey Data 84
B Baggage Heat-Release Rates 92
C FDS Simulation Results 99
Bibliography 105
iv
List of Tables
1.1 Sampling of ferry fires in the last 25 years. . . . . . . . . . . . . . . . 5
2.1 Comparison of density and thermal conductivity of common steel and
aluminum alloys [1, 2]. . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Comparison of USCG Fire Load Requirements from oldest to most
recent. The CFR restricts fire load based on a low fire load accom-
modation space. With the introduction of PFM 1-94 the remaining
requirements are related to the newly created Type 5A space with
very low fire load. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3 Summary of the relevant aspects of NVIC 9-97 Change 1 requirements
for compartments to utilize a type 5A very low fire load exemption. . 24
2.4 Comparison of USCG seat cushion data from testing performed in
1998 and 2012. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.5 Relevant test dat from Swedish Bag Tests [3]. . . . . . . . . . . . . . 34
3.1 Seat density and transient fire load for three ferry vessels from which
carriage data was collected. . . . . . . . . . . . . . . . . . . . . . . . 47
v
List of Figures
1.1 Burned remains of the LEVINA I ferry, a disaster that resulted in
the deaths of more than 40 passengers [4]. . . . . . . . . . . . . . . . 10
2.1 Setup of the test compartment aboard the STATE OF MAINE [5]. . 19
2.2 Picture of the M/V IYANOUGH the subject vessel of the 2012 study [6]. 26
2.3 Top-view showing the general arrangement of a vessel with a similiar
layout to the M/V IYANOUGH [7]. . . . . . . . . . . . . . . . . . . . 27
2.4 Picture of unburned rail-cars that were abandoned during the Baku
metro fire [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.1 Allowed weight per passenger based on seat density and using a con-
servative full load assumption. The red dashed lines bound the nor-
mal range of seat densities found in currently operating ferry vessels. 38
3.2 Representative samples of baggage that were weighed on commuter
and non-commuter ferries. . . . . . . . . . . . . . . . . . . . . . . . . 44
3.3 Histogram showing the weight of baggage and cumulative percentage
carried aboard ferry vessels. . . . . . . . . . . . . . . . . . . . . . . . 45
4.1 Experimental measurements and ramp functions for middle and side
ignition of seats given by Shriner [8]. . . . . . . . . . . . . . . . . . . 50
4.2 Grid resolution comparison showing the average ceiling temperature
across the passenger compartment for three different grid resolution
values. The ceiling temperature (the important parameter for this
study) converges at a grid resolution value of 10. . . . . . . . . . . . . 54
4.3 Weighted heat release rate curve for an average bag carried aboard a
ferry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.4 FDS setup for burning of the stroller in the corner. . . . . . . . . . . 60
4.5 FDS setup for burning of two seat banks. The burning carry-on bag
is shown by the orange box in the middle, with the seat cushions
shown in blue. Heat flux gauges for the left hand seats are visible as
two green dots, which are oriented to face the fire. . . . . . . . . . . . 62
vi
4.6 FDS setup for burning of a single seat bank. The initial seats set
on fire are indicated as orange seats, with heat flux gauges shown as
green dots. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.1 Corner ceiling temperature above burning stroller. . . . . . . . . . . . 65
5.2 Heat-release rate from a simulation of burning of two seat banks. . . . 68
5.3 Ceiling temperature profile above two burning seat banks at 1434
seconds after ignition of the carry-on bag. . . . . . . . . . . . . . . . 69
5.4 Average temperature over the square meter of ceiling above the lower
four burning seats. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.5 Heat release rate from a single bank of burning seats with baggage. . 72
5.6 Average ceiling temperature over one square meter above burning
single seat bank. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
A.1 Baggage carriage distribution for ferry vessels. . . . . . . . . . . . . . 85
A.2 Baggage carriage distribution including only small bags carried into
the passenger compartment. . . . . . . . . . . . . . . . . . . . . . . . 85
A.3 Baggage weight histogram for all baggage combined. . . . . . . . . . . 86
A.4 Baggage weight histogram for purse. . . . . . . . . . . . . . . . . . . 86
A.5 Baggage weight histogram for backpack. . . . . . . . . . . . . . . . . 87
A.6 Baggage weight histogram for small duffel bag. . . . . . . . . . . . . . 87
A.7 Baggage weight histogram for large duffel bag. . . . . . . . . . . . . . 88
A.8 Baggage weight histogram for carry-on size suitcase. . . . . . . . . . . 88
A.9 Baggage weight histogram for briefcase. . . . . . . . . . . . . . . . . . 89
A.10 Baggage weight histogram for tote bag. . . . . . . . . . . . . . . . . . 89
A.11 Average weights of each specified baggage type. . . . . . . . . . . . . 90
A.12 Comparison between average weights from Swedish Study [3] and
vessel survey data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
A.13 Measured weight data from all baggage weighed during vessel surveys. 91
B.1 Liner heat-release rate curve use for a stroller based on data from
Kumm [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
B.2 Liner heat-release rate curve use for a carry-on bag based on data
from Kumm [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
B.3 Liner heat-release rate curve use for a briefcase based on data from
Kumm [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
B.4 Liner heat-release rate curve use for a purse based on data from
Kumm [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
B.5 Liner heat-release rate curve use for a small duffel bag based on data
from Kumm [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
B.6 Liner heat-release rate curve use for a backpack based on data from
Kumm [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
C.1 Single point ceiling temperature above the burning stoller. . . . . . . 100
C.2 Ceiling temperature above burning stroller in the corner of the com-
partment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
vii
C.3 Maximum single point ceiling temperature above a single bank of
burning seats. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
C.4 Maximum single point ceiling temperature above two burning seat
banks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
C.5 Average ceiling temperature over one square meter above two burning
seat banks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
viii
List of Abbreviations
PFM Policy Fire Memorandum
MTN Marine Safety Center Technical Note
FDS Fire Dynamics Simulator
USCG United States Coast Guard
MSC Maringe Safety Center
NVIC Navigation and Vessel Inspection Circular
CFR Code of Federal Regulations
CFD Computational Fluid Dynamics
FDS Fire Dynamics Simulator
NIST National Institute of Standards and Technology
PVA Passenger Vessel Association
HRRPUA Heat-Release Rate Per Unit Area
NTSB National Transportation Safety Board
ix
Chapter 1: Introduction
1.1 Regulatory History
Federal requirements governing the amount of flammable material brought
aboard commercial passenger vessels have been in place for many years. These
requirements are based loosely on information that dates back to the time of the
first cruise ship type passenger vessels. A number of studies in the last 20 years
have provided data and information on the appropriateness of current regulatory
requirements, but this data is not complete. Although the interior furnishings such
as tables, seats, carpeting, and veneers have been studied closely, the transient fire
load from baggage has been largely ignored with the exception of somewhat arbitrary
regulations restricting the weight based solely on the area of the passenger deck.
The prevalent use of aluminum as a building material beginning in the 1980s
led the United States Coast Guard to enact special requirements for structural fire
protection of aluminum vessels, but may place burdensome roles on vessel operators.
Data is needed to ensure that Coast Guard requirements are appropriate and that
vessel operators have a clear understanding of how to enforce regulations aboard
their vessels.
1
1.2 Approach
The first step in order to address this problem is to determine whether or
not current regulations are being met with regard to weight of transient fire load
combustibles brought on board vessels. If the current regulation is not being met,
the total extent of transient fire load being brought on board must be determined.
Aluminum ferry vessels will be selected for audit and the baggage taken aboard will
be recorded. The number of bags brought aboard, along with the baggage type will
be recorded. This data will allow the formation of an overall carriage rate as well
as determine what types of bags are most prevalent for carriage. Next the baggage
brought aboard will be weighed to determine its contribution to the transient fire
load. The weights will also be recorded based on baggage type to determine if
certain bags are on average too heavy to be brought aboard. This will also allow for
an average weight to be calculated for the purpose of knowing the average transient
fuel load passengers bring aboard.
The next step is to create a computer simulation of the main passenger com-
partment of the ferry. The computational fluid dynamics (CFD) program Fire Dy-
namics Simulator (FDS) created by the National Institute of Standards and Tech-
nology (NIST) will be used to model the compartment and the fire to determine
the heating of the aluminum overhead. A generic passenger compartment is created
using overall dimensions of an in-service ferry vessel in Massachusetts. The seats
will be modeled as rectangular objects with a prescribed heat release rate per unit
area based on extensive full scale seat burning tests completed in 2012. This testing
2
also revealed that seats are the only significant contributors to the fixed fire load
within very low fire load passenger compartments. Data collected on the carriage
and weight of the baggage will be coupled with baggage burn data from tests com-
pleted in 2010. Multiple scenarios will be considered, including the effect of baggage
placed in corners and along walls, as well as multiple bags on a seat bank and bags
placed on a table top. From the considered scenarios three will be selected for full
scale simulation using FDS.
Simulations will be repeated multiple times to predict the spread of the fire
between seats, seat banks, and baggage. The focus will be on creating conservative
scenarios, meaning that each scenario will focus on a potentially dangerous config-
urations given the initial circumstances of the scenario. The scenarios will serve as
an aid in determining if the specific placement of baggage, or the sheer volume of
baggage in and around seats can possibly create an atmosphere within the compart-
ment that will raise the temperature of the aluminum overhead to a temperature at
which the structural integrity is compromised.
1.3 Objective
The goal of this study is twofold: first, to ascertain the current state of regu-
lations regarding the carriage of baggage, referred to as the transient fire load, on
aluminum ferry vessels; second, to provide a comparative analysis of the addition
of transient fire loads in order to assist in measuring the risk associated with the
current rate of carriage.
3
The current requirements in place for vessel operators allow for a fixed weight
of baggage to be brought aboard for each square meter of passenger space. Although
this figure is useful from a calculation perspective, it does not provide a tangible
metric for use in regulating what passengers bring aboard the vessel. There is a
lack of data regarding what amount and type of baggage constitutes the required
maximum load of 2.5 kg/m2 (0.5 lb/ft2).
1.4 Past Fires
Although there have not been any significant fires in passenger compartments
on day ferries, it is obvious from past experience that the unexpected often occurs.
In today?s social climate, with the threat of terrorist activities constant, and with
the ever-present possibility of electronic and mechanical failure, the potential for a
fire ignition in a passenger compartment cannot be ignored. It is the intent of this
study to ensure that the conditions aboard these heavily-used day ferries be studied
and modeled to the extent that the risk of a fire event that has a high casualty
rate be minimized to the greatest extent possible. This will be done by ensuring
that realistic fire loads are maintained aboard ferries, and by examining a range of
probable fire and compartment layout scenarios.
Table 1.1 shows a sample of fires that have occurred on ferries within the last
quarter century. Although the majority of fires have resulted in a zero death toll,
one fire in 2006 resulted in the deaths of more than 1,000 people. This highlights the
very real fire risk that is still present in modern day ferry vessels. The list of fires
4
Table 1.1: Sampling of ferry fires in the last 25 years.
Vessel Date No. of Deaths Fire Origin
CATALINA ISLAND FERRY 25-Jan-89 0 Engine Room
SCANDANAVIAN STAR 07-Apr-90 159 Passageway
PRINCESS RAGNHILD 08-Jul-99 0 Machine Room
SUPERFERRY 14 26-Feb-04 1 Engine Room
AL SALAM BOCCACCIO 03-Feb-06 1,031 Car Deck
LEVINA I 22-Feb-07 41 Car Deck
LADY MARTHA 18-Sep-08 0 Engine Room
LISCO GLORIA 09-Oct-10 0 Upper Deck
PELLA 03-Nov-11 1 Car Deck
ANGELINA LAURA 09-Apr-79 0 Galley
5
that resulted in zero deaths appear to be somewhat passive events because of this
statistic, but injuries and smoke inhalation are not reported and may be significant
in some of these scenarios. It is important to note that many of these fires occurred
overseas and not domestically. The United States is certainly a leading country
in terms of the safety of its seafaring community, but until fire has been all but
eliminated as a waterborne transportation risk it bears further study to approach
the most efficient and cost effective mitigation methods.
1.4.1 Scandanavian Star
The first vessel fire relevant to this study is a blaze that occurred on board the
M/V SCANDANAVIAN STAR. The SCANDANVIAN STAR was a 141 meter roll
on roll off ferry that was capable of carrying 1152 passengers and 100 crew members
along with a full load of cars and trucks. On the night of April 7, 1990 the ferry was
making an overnight crossing from Denmark to Norway carrying 383 passengers and
99 crew members [9]. At 1:55 am a fire was started outside cabin 416 by an arsonist,
but was quickly discovered and extinguished. At 2:00 am another fire was started
by an arsonist outside cabin 219. This fire spread up through the ladder well of the
ship and pushed smoke out into the passageways on decks 4 and 5. A number of
open fire doors allowed the smoke to travel quickly and the fire to spread rapidly.
In addition, the ventilation system was adjusted to a maximum flow which fed the
fire a supply of fresh air and increased its spread rate. Over the course of the day
a total of 4 more fires were started, but none with the significance and impact of
6
the second fire. The rapid spread of the fire and large amounts of smoke hindered
evacuation and a total of 159 people perished in the fire [9].
The fire on the SCANDANAVIAN STAR shows that human factors such as
leaving open fire doors or making negative adjustments to the ventilation system,
whether by ill intent or accident, can have a significant effect on the spread of a
potential fire and tenability of compartments. In addition, it is a reminder that
even in the absence of a high ignition risk equipment, fires can sometimes be started
by people in unexpected areas and quickly get out of control. The risk of fire from
arson or a terrorist attack is a real danger in the modern world and cannot be ruled
out as a possibility when considering fire protection engineering for mass-transit
systems such as ferries.
1.4.2 Al Salam Boccaccio 98
The second fire of importance for this study is the fire that resulted in the
sinking of the M/V AL SALAM BOCCACCIO 98. The AL SALAM BOCCACCIO
98 was a 131 meter roll on roll off ferry that was capable of carrying more than 1,300
passengers, 100 crew and a full load of vehicles. On the evening of February 2, 2006
the AL SALAM BOCCACCIO 98 was making an overnight crossing from Saudi
Arabia to Egypt carrying 1, 321 passengers, 97 crew members, 22 cars, 14 trucks,
and 7 trailers [10]. At 7:09 pm the fire alarm on the AL SALAM BOCCACCIO
98 activated and one minute later the watchmen from the car deck arrived at the
bridge to report that the car deck was full of black smoke and he believed the fire
7
was coming from the engine room. The master ordered the activation of the water
spray system in the car deck and ordered that fire hoses and teams be sent down
to combat the fire. Several other reports of smoke and fire came in to the bridge
and the master ordered water to mitigate the smoke in the cabin areas. At 7:36 pm
the location of the fire was finally identified as the luggage trailer in the forward
port side car deck. The master of the vessel continued to send hose teams to spray
water anywhere that smoke was detected, even if no fire was present. The fire was
fought for about 30 minutes before pumping of the fire water began to clear the car
deck, however the list of the ship, caused by mostly unnecessary suppression efforts,
was causing water being pumped out to return to the starboard side. The list of
the vessel increased despite efforts to de-water and by 11:30 pm the list was at 25
degrees; 3 minutes later the vessel sank [10].
The fire on the AL SALAM BOCCACCIO 98 demonstrates that even with
detection and suppression systems in place, sometimes conditions quickly get to a
point beyond control. The AL SALAM BOCCACCIO 98 was not a day ferry as are
the ferries that are the focus of this study, but it is an example of how a fire even
in a supposedly well controlled space can quickly overcome installed systems and
create insurmountable problems.
1.4.3 Lady Martha
The final fire of importance for this discussion was an incident aboard the
LADY MARTHA. This vessel is a ferry that takes passengers from Cape Cod, MA
8
to the island of Nantucket, MA. The fire started in the starboard engine room of
this vessel just prior to arrival in port [11]. The fire could easily have spread to the
passenger compartment if it had gotten out of control. The fire fighting system in
the engine room activated, knocking down the fire, and the vessel was able to dock
at the pier despite the smoke coming out of the vessel. The ferry only had about 6
people on board, so these people quickly disembarked and the swift notification from
the harbor master ensured that shore-side firefighting personnel were on there way.
The shore-side firefighting teams entered the vessel and completed extinguishment
of the fire [12].
1.4.4 Summary
In all cases reviewed, incidents occurred on passenger ferries where overnight
accommodations were available or did not involve passenger areas. Despite a lack
of recorded incidents, the risk exists for a costly fire to occur in the passenger
compartment of a day ferry. Several close calls, such as the LADY MARTHA fire,
support the potential risk of more intense fires with greater casualties.
Many times regulatory changes and policy adjustments are made as a reac-
tionary step following a disaster, however the purpose of this study is to ensure
that information is provided that allows changes to be made prior to a disaster in
a effort to prevent a loss. The fire on the AL SALAM BOCCACCIO showed that
even with modern day suppression systems in place fires can quickly overwhelm
systems. The fire on the LADY MARTHA shows that high speed ferries, which are
9
Figure 1.1: Burned remains of the LEVINA I ferry, a disaster that re-
sulted in the deaths of more than 40 passengers [4].
the focus of this study, are also susceptible to fires and even with proper action of
the suppression system the shore-side fire department was required to provide final
extinguishment of the fire. Finally, the SCANDANAVIAN STAR tragedy shows
us that even when mechanical or accidental means are ruled out, there is still the
possibility of fire ignition. In today?s unstable world where terrorism and crime are
constant threats, the possibility of arson or a terrorist attack leading to a fire are
very real. It must be ensured that should a fire begin in a passenger compartment
or spread to a passenger compartment from another space, the the outfitting and
luggage carriage of the vessel will not allow the passenger compartment fire to reach
a critical level that would prevent passengers and crew from evacuating the vessel.
10
Chapter 2: Literature Review
2.1 Vessel Inspection History
The Federal Government has been involved in the regulation of commercial
vessel safety since 1838 when Congress passed a law requiring better security of
the lives of passengers aboard steam vessels. The primary safety concern in the
beginning was explosions and fire caused by vessel machinery. Despite the passage
of several laws over the years, steamboat disasters continued to be such a large
problem that, in 1871 the Steamboat Inspection Service was created to regulate the
safety of these vessels. This service was later shifted to the control of the US Coast
Guard (USCG) and regulations have continued to adapt with time and technology.
Current safety requirements are set out in the Code of Federal Regulations (CFR)
as specified by United States law. Policy documents such as Navigation and Vessel
Inspection Circulars (NVIC) and Marine Safety Center Technical Notes (MTN)
clarify the regulations and offer guidance on acceptable ways to meet the regulations.
Shipboard fire scenarios are extremely dangerous because immediate evacua-
tion is not possible and fire fighting options are severely limited. Unlike a building,
evacuation is not a simple process of walking out a door or down some steps; to
evacuate a vessel at sea you must first locate and don a life preserver, wait while the
11
crew launches lifesaving devices, and carefully embark the lifesaving device, often
times by climbing down a ladder or net in order to board the raft or lifeboat. In
addition, because ships must maintain stability and buoyancy, the amount of wa-
ter used to combat fires must be carefully limited, and pumping operations create
additional tasks for the crew when battling a large blaze. All of these factors in
combination mean that in order to ensure the safety of passengers during a fire,
there must be a safe and secure location where the passengers can muster, don their
life preservers, and embark lifesaving devices.
In addition to detailed regulations regarding detection and suppression, cur-
rent regulations also require that large vessels must have refuge areas adequately
protected from smoke, fire, and heat for a time period adequate for the evacuation of
passengers and crew. Structural insulation has long been a heavily utilized method
of fire protection that allows for the containment of fire, smoke, and heat for a pe-
riod of time sufficient for evacuation of the vessel to occur. Insulation can contain
the fire to the compartment of origin, retard the spread to other compartments, and
provide protection for refuge areas. Adequate fire insulation along with noncom-
bustible materials with large thermal inertia has long been the preferred method to
ensure adequate evacuation time. Steel has also long been the material of choice
for shipbuilders, and a great many vessels are constructed with this material. Con-
sequently the regulations were written primarily with steel in mind. However, in
the 1980?s aluminum was becoming an increasingly popular building material for
commercial vessels because its light weight meant higher speeds and lower fuel costs
compared with similar steel vessels.
12
The level and degree of insulation required for a space is based on the use and
location of the space. Spaces with an increased risk of ignition, such as machinery
spaces, are required to have a high degree of insulation. Compartments bordering
these high risk spaces must also have adequate boundary insulation. Compartments
with very low risk of ignition are required minimal insulation, with uninsulated steel
potentially being an adequate barrier surrounding the compartment. The regula-
tions were written with the thermal properties of steel in mind, but a growing interest
in aluminum vessels would eventually necessitate a change to the requirements.
2.1.1 Aluminum Vessels
In the early 1980?s shipyards and naval architects began pushing to create a
new kind of passenger ferry. In order to increase the number of passengers moved,
builders and owners wanted to create high-speed ferries that could complete tradi-
tional ferry runs in as little as half the time. Aluminum was the material of choice
for the design of this new type of vessel because the reduced weight of the hull
and superstructure allowed the vessels to maintain a dead-weight low enough to be
able to achieve high speeds without the need for massive engines. When it came
to structural fire protection the designers and engineers ran into a problem, the
difference in thermal properties between aluminum and steel meant that the alu-
minum bulkheads in these new vessels would have to be heavily insulated in order
to meet the minimum requirements. This additional insulation would effectively
negate the weight savings of using aluminum and eliminate the advantages of the
13
design. The high density and low thermal conductivity of steel meant that it had a
greater thermal capacity to absorb heat than an equal thickness of aluminum and
the steel would not transfer the heat to the non-fire side of the bulkhead as quickly.
Table 2.1 shows the differences in properties between aluminum and steel.
Table 2.1: Comparison of density and thermal conductivity of common
steel and aluminum alloys [1, 2].
Material Density (kg/m3) Thermal Conductivity (W/mK)
Aluminum, 6082 2700 180
0.4% Carbon Steel 7850 48
USCG regulations promulgated in Title 46 of the Code of Federal Regulations
(CFR) specify the requirements for insulation of decks, bulkheads, and overheads.
These requirements are based on insulating to a level that will prevent a temperature
rise on the unexposed side over 121?C (250?F) during the hour-long standard fire
test. The standard fire test is defined in 46 CFR 114.400 as an exposure based on
a curve that should pass through five specified temperature points in time. The
requirement for insulation is based on the location and use of the space. When
considering space usage as a risk factor, both ignition probability and fire load are
contributors to the risk level a space presents. Fire load is a term used to describe
the weight of all combustibles in the space normalized by the area of the space.
In order to alleviate the problem created by the difference in thermal properties
14
between steel and aluminum, designers appealed to the argument of low fire load.
In 46 CFR 116.427 the regulations specify that low risk accommodation spaces must
have fire load calculations completed. Fire load calculations are done to ensure that
the density of combustible material is below a prescribed value. The calculations
consist of taking the weight of every combustible item in the accommodation space
and dividing this number by the area of the deck. The maximum fire load for a low
risk accommodation space has been defined as 15 kg/m2 (3 lbs/ft2).
Several shipbuilders teamed up to create a plan whereby they would eliminate
structural fire protection insulation in passenger compartments of large aluminum
vessels while concurrently reducing the fire load to lower the risk of fire and reduce
the hazard should a fire start. The proposal centered around the idea that, should
a fire start in one of these low fire load spaces, the lack of available fuel in the
space would keep temperatures low enough that the criteria for temperature rise
would be met by aluminum bulkheads and decks without the need for fire protective
insulation. Since this plan did not adhere to the current regulations, the ship builders
had to appeal directly to the USCG Marine Safety Center (MSC) in Washington,
D.C. to get approval for this equivalent safety design. The USCG MSC decided that
the plan presented by designers and shipbuilders did achieve an equivalent level of
safety to that of the current regulations and approval of designs was granted on a
case by case basis for several high speed aluminum vessels.
In the early 1990?s other shipbuilders began taking a serious interest in building
aluminum high-speed vessels in order to take advantage of the same equivalent safety
design utilized by the few that had pioneered the design. At the time the USCG
15
MSC was still only approving this equivalent safety design on a case by case basis.
As industry pressure began to mount, the USCG sought to codify the equivalencies
presented in the equivalent safety design cases, and in 1994 they published Policy
Fire Memorandum (PFM) 1-94. PFM 1-94 created a policy whereby any shipbuilder
could eliminate the structural fire protection insulation requirement by ensuring that
the vessels had an appropriately low fire load in the passenger compartment, and
met stricter requirements for interior materials and fire detection [13].
PFM 1-94 officially created the concept of very low fire load accommodation
spaces and designated these spaces as Type 5A. Previous exemptions were given
on a case by case basis, but this policy memorandum synchronized all of the re-
quirements to ensure that all vessels using the structural fire protection insulation
exemption met certain conditions. PFM 1-94 limited Type 5A spaces to vessels
carrying not more than 600 passengers, with no overnight accommodations, and
with predominately open public spaces with a uniformly distributed fire load. The
fire load for these spaces was limited to 5.0 kg/m2 (1.0 lb/ft2), and they were re-
quired to be constructed with noncombustible trim and veneers and fire resistant
furnishings [13].
PFM 1-94 allowed a C-Class, smoke-tight structural boundary of aluminum
(without insulation) to be used where a refuge area was located adjacent to the
5A space. This meant that even if a refuge area was designated for the bow or
upper deck, bare aluminum could be used as long as the fire load was low. This
allowed designers and builders to create vessels with significantly reduced weight,
particularly in the superstructure of the vessel.
16
2.1.2 Coast Guard Research
With the number of aluminum vessels rapidly increasing the USCG decided to
validate the assumption that type 5A spaces had sufficiently low fire load in order
to maintain temperatures low enough to meet the criteria for temperature rise on
the unexposed side. The primary tool for evaluating this assumption would be a
test to ascertain whether or not a compartment with low fire load would flashover
prior to the evacuation of all persons. The full-scale tests were conducted by USCG
Research and Development Center personnel aboard the vessel STATE OF MAINE.
The test compartment had dimensions of 4.6 m wide, by 5.0 m deep, by 2.5 m high
and was constructed of steel. A 6 mm thick aluminum drop ceiling was positioned
0.1 m below the overhead. The test utilized seats that provided a fire load of 5.0
kg/m2 (1.0 lb/ft2) arranged in 3 rows of 9 chairs and one row of 8 chairs. The seats
were ignited using a 15 cm or 10 cm diameter pan of heptane that was 3 cm deep.
Tests were done with different seat configurations and with the addition of wood
cribs for one test to increase the fire load. The combustible weight of the seats alone,
spread out over the entire 23 m2 deck area only equated to a total fire load of 1.52
kg/m2 (0.31 lb/ft2), well short of the 5.0 kg/m2 (1.0 lb/ft2) fire load limit given by
the regulations. With the addition of 24 wood cribs, each weighing 5.0 kg, the fire
load in the space was raised to 6.74 kg/m2 (1.38 lb/ft2), which easily exceeds the
requirement [5].
Heat-release rates for the tests were measured using oxygen consumption
calorimetry with concentrations and flow rate measurements being taken at the
17
doorway and integrated over the area above or below the neutral plane, dependent
on whether inflow or outflow was being used. The flow to be used was determined by
comparing the flow rate based on a heat of combustion for the seats of 21,000 kJ/kg
and 13,000 kJ/kg for the wood cribs. The two tests that utilized the wood cribs,
and thereby achieved the correct fire load density, achieved peak heat-release rates
of 560 kW and 370 kW dependent on the arrangement of the seating. In addition
to heat-release rates, the temperature on the interior and exterior of the aluminum
panel ceiling were measured. The measurements were taken with 5 Type-K thermo-
couples sheathed with Inconel and peened into the aluminum panel at five different
locations. The peak ceiling temperature for the test completed with the additional
wood cribs was 385?C on the interior and 247?C on the exterior of the aluminum
panel [5].
The final test done as part of the USCG full scale testing series was to place
the seat cushions from all 35 seats in one massive pile in the center of the room.
Although this still equated to a relatively low fire load over the whole space (still
only 1.52 kg/m2), the proximity of the cushions to one another created a large fire.
This fire had a peak heat release rate of 2200 kW and achieved ceiling temperatures
of 657?C [5]. The purpose of this test was to investigate the effect that fuel loading
and configuration had on the heat-release rate and to determine whether a worst-
case scenario, such as the stacking of the seat cushions in the center, was sufficient
to cause flashover in the compartment.
In addition the test investigated the temperature that aluminum panels would
reach when exposed to different fire scenarios. The aluminum temperatures mea-
18
Figure 2.1: Setup of the test compartment aboard the STATE OF MAINE [5].
19
sured did not conclusively reach the melting temperature of aluminum, but far
exceeded the 232?C temperature which corresponds to a 50% loss of structural in-
tegrity for aluminum. In order to investigate the impact of flame impingement on
the ceiling of the bulkhead of an aluminum compartment, a series of tests were
done with a 61 cm by 61 cm section of the 6 mm thick aluminum ceiling panel.
The tests showed that a heat-release rate of 25 kW caused a panel temperature of
300?C and 60 kW caused a temperature of 400?C. Using this test data, along with
the START*CD CFD program and correlations for fires against a wall, the authors
of the test concluded that an estimated fire size of 180 kW positioned next to a
bulkhead would cause the aluminum to reach its melting point after 10 minutes.
The remainder of the test was a comparison of the full scale experiments with the
USCG fire modeling program SAFE. The results of this test could prove significant
in that, baggage placed against a bulkhead could easily create a fire of 180 kW and
thereby compromise the structure of an aluminum bulkhead within 10 minutes.
Following the full scale testing and verification of current assumptions aboard
the STATE OF MAINE, the USCG promulgated NVIC 9-97 on Structural Fire
Protection, including chapter 4 which dealt with aluminum vessels. Chapter 4 of
NVIC 9-97 concerns the structural fire protection guidelines for aluminum vessels,
and section 4.2 speaks specifically to very low fire load spaces of concern for this
research. In order to qualify as a very low fire load (Type 5A) space it must be a
large open public space with uniformly distributed fire load not to exceed 5.0 kg/m2
(1.0 lb/ft2), must have fire resistant furnishings and finishes, must have an approved
fire detection and manual fire alarm system, and an A-II portable fire extinguisher
20
Table 2.2: Comparison of USCG Fire Load Requirements from oldest
to most recent. The CFR restricts fire load based on a low fire load ac-
commodation space. With the introduction of PFM 1-94 the remaining
requirements are related to the newly created Type 5A space with very
low fire load.
USCG Requirement Year Fire Load (kg/m2) Transient Fire Load (kg/m2)
PFM 1-94 1994 5.0 N/A
CFR 1996 15.0 N/A
NVIC 9-97 1997 5.0 N/A
NVIC 9-97 Change 1 2010 5.0 0.75
MTN No. 01-13 2013 5.0 2.5
must be provided for every 45 m2 of deck area. The fire load for passenger effects
is not to exceed 0.75 kg/m2 (0.15 lb/ft2) [14]. There are several other requirements
detailing the construction materials and special conditions that can be found in
NVIC 9-97 4.2.1. The requirement for passenger effects was new, and introduced a
further variable in maintaining a low fire load, one that must be carefully managed
by vessel operators. Table 2.2 shows the progression of the regulations from the
first policy in 1994 to the most recent policy given in Marine Technical Note (MTN)
01-13.
21
2.1.3 Regulatory Changes
In the late 2000?s the fire load issue was again re-examined to ensure the policy
provided an equivalent level of safety to the full regulations. Advances in computer
technology since the testing done in 1997 made it possible to conduct Computational
Fluid Dynamics (CFD) modeling. The modeling conducted in conjunction with
this project showed that the safety of the area of refuge, when no insulation was
present, was highly dependent on the type and distribution of the seating and other
combustible materials within the space as well as the ventilation conditions. The
dependency of fire size and severity on the configuration of combustible materials
had been clearly shown by the compartment testing completed by the USCG in
1997, but the new modeling work suggested that the dependency on fuel load and
configuration was more severe than previously thought. Unfortunately, an official
report was not published for this work, and specific details on the modeling are not
currently available. The conclusions drawn from this modeling effort presented a
large dilemma because there were numerous vessels in service taking advantage of
the type 5A insulation exemption. These vessels all had sufficiently low fire load
to meet NVIC 9-97, but the new research suggested that some of these vessels may
be at much higher risk due to the configuration of combustibles on board and the
status of ventilation.
The only way to know which vessels were truly low risk and which were in
potential danger, was for vessel operators to evaluate fire load and configurations
for each vessel. In 2010 the Coast Guard promulgated Change 1 to NVIC 9-97
22
which now required that boundaries between Type 5A spaces and areas of refuge be
insulated to achieve a fire rating of A-60. This meant that the boundary must be
capable of preventing the passage of smoke and flame for one hour as well as prevent
the temperature of the unexposed side from rising by more than 131?C (250?F) on
average or 181?C (325?F) at any point after 60 minutes. Meeting these requirements
for an aluminum vessel meant following prescriptive insulation requirements which,
in addition to costing thousands of dollars per vessel, would cripple the speed of the
vessels by adding a massive amount of weight. If a vessel wanted to keep its current
construction without any insulation it must submit a full engineering analysis to the
Coast Guard to prove that the area of refuge would be safe for the full evacuation
time.
Aluminum vessel builders as well as owner/operators who had been taking
advantage of the Type 5A policy given in NVIC 9-97 were now in the position of
paying thousands of dollars for each of their vessels to either install insulation or have
an engineering analysis performed. Several industry representatives got together
along with the Passenger Vessel Association (PVA) and asked the Coast Guard to
create a method for compliance that would allow vessel builders and owner/operators
to comply with the new policy without spending thousands of dollars per vessel. In
response to this request the Coast Guard worked with a University of Maryland
(UMD) graduate student to study the problem and create a potential solution for
vessel owners. The result of the research and interaction between industry and the
Coast Guard was the 5A Space Performance Guidelines which are published in MTN
01-13.
23
Table 2.3: Summary of the relevant aspects of NVIC 9-97 Change 1
requirements for compartments to utilize a type 5A very low fire load
exemption.
1 Large open public space with a uniformly distributed fire load.
2 Fire load calculations showing 5.0 kg/m2 fire loading.
3 Interior finishes must meet 46 CFR 164.012 or .112.
4 Furniture, draperies, and carpets must be fire resistant.
5 Fire detection and manual fire alarm system must be installed.
6 One type A-II portable fire extinguisher for every 45 m2.
24
2.2 Recent Work
The Coast Guard utilized its graduate student program to study the issue of
the fire load in aluminum ferries, in order to try and create a pre-approval system
whereby vessel builders would be able to follow a simple set of rules in the layout and
construction of type 5A spaces in order to make full use of insulation exemptions.
Creating a simple set of rules would allow vessels to continue to operate effectively
without structural insulation in the superstructure, while maintaining an approriate
level of safety.
2.2.1 Coast Guard FDS Modeling
The Coast Guard, PVA, Nichols Brothers Boat Builders, and Gladding-Hearn
Shipbuilding worked together utilizing a study done by UMD student Noel Shriner
to refine the policy put out through NVIC 9-97 Change 1 regarding fire loads in very
low risk accommodation spaces. The purpose of the study and subsequent policy
document was to identify a pre-approved arrangement that could be accepted in lieu
of a full engineering analysis. This would allow operators to take advantage of the
exemption policy without spending large amounts of money insulating passenger
spaces or conducting individual studies for each vessel.
This study was conducted with the purpose of validating current regulations
related to the maximum fire load in very low risk accommodation spaces, and creat-
ing seating configuration rules that could be utilized by industry to take advantage
of the insulation exemption without doing a full engineering analysis. The structural
25
Figure 2.2: Picture of the M/V IYANOUGH the subject vessel of the
2012 study [6].
integrity of the overhead above a burning compartment was evaluated to ensure that
the temperature of the aluminum overhead did not exceed a certain level deemed
critical. At 232?C aluminum is considered to have lost half of its structural integrity,
therefore a value of 232?C was used as a threshold temperature; any values greater
than this would be of serious concern.
The 2012 study built on past research by using current CFD modeling tools to
evaluate a modern aluminum ferry passenger compartment. The study was modeled
after the M/V IYANOUGH, representative of a typical vessel with Type 5A spaces.
The layout of the vessel?s lower passenger compartment was modeled using Fire Dy-
namics Simulator (FDS) Version 5, a large eddy simulation (LES) computer model
developed by the National Institute of Standards and Technology (NIST) [15]. The
study considered flame spread between seats, the influence of the spacing of seats,
26
Figure 2.3: Top-view showing the general arrangement of a vessel with
a similiar layout to the M/V IYANOUGH [7].
and the overall heat release from an estimated worst case scenario. Performance cri-
teria included requirements that the deck above not reach an average temperature
of 200?C over any square meter and that no point on the deck reach 400?C.
In order to ensure that the fire was represented appropriately all possible
combustible materials had to be considered. These included tables, chairs, and
carpet. The tables in the study, however were considered noncombustible and were
thus ruled out as a contributor to the fire load.
The carpet was tested to determine ignition and flame spread properties. In-
stalled carpet must meet the ASTM E84 (116.423). The carpet was tested with a
simple butane lighter, but failed to sustain a flame. Next the carpet was subjected
to flame tests after having a portion soaked with heptane fuel. The carpet was not
able to sustain flaming combustion even with a section soaked in 150 mL of heptane.
The section of carpet that had the heptane burning was damaged, but the carpet
itself did not sustain a flame and did not spread the flame laterally. The carpeting
27
was therefore ruled out as a contributor to the fire load in the space.
Next the seats were tested for their flammability properties. The seats were
required to meet the fire resistance standard UL 1056 in order to be approved for
type 5A spaces. The seats have a combustible weight of 1.6 kg and a plan view area
of 0.2 m2. This creates a localized fire load of 7.0 kg/m2 (1.43 lb/ft2), but distributed
over the entire space studied this equates to a fire load of 1.27 kg/m2 (0.26 lb/ft2).
The foam used in the seats varies between the seat cushion, seat back, and other
foams to add stiffness. The most predominant foam, however is EN 38-200 which
has a density of 38 kg/m3. This foam was assumed to account for the entire 1.6 kg
of combustible weight.
Initial testing on the seat cushions showed that the fabric was capable of
sustaining a flame long enough to melt the foam and establish a small pool fire on the
melted foam. Samples of the seat cushion were then burned in the cone calorimeter
to determine the critical heat flux, time to ignition, and heat of combustion. The
critical heat flux was 10 kW/m2 and the heat of combustion of the seat cushion was
found to be 17.3 MJ/kg. Next, the seats were tested for their overall heat-release
rates. Two ignition scenarios were tested, ignition underneath the middle of the
seat, and side ignition from an adjacent seat. The peak heat-release rate measured
from both of the ignition scenarios was 100 kW ?3kW.
The area under the heat-release rate curves was calculated and linear trend
lines were used to represent the curves in a more simple fashion. Ensuring that the
area under the trend line curve was equal to that of the measured heat-release rate
curve ensured that the total heat released was conserved. The total heat released
28
for the burning chairs was 22.7 MJ for the middle ignition scenario and 21.8 MJ for
the side ignition scenario.
The chair was then modeled as a burner within FDS, using a specified burning
rate according to the trend-line heat-release rate curve from the testing. The burner
was placed at a height of 0.5 m, representing the bottom seat cushion because the
burning of the seat was assumed to mostly occur on the bottom cushion. The
main passenger compartment of the M/V IYANOUGH was modeled in FDS with
each seat representing a burner. The burners were programmed to begin releasing
heat at a prescribed time based on the time to ignition for side ignition of the seat.
Several adjustments and corrections were made to the model, inter alia, changing the
radiative fraction to correspond with the values measured during full-scale testing.
Once all the corrections and adjustments were made, the model was run to
determine the temperature of the aluminum overhead for bare aluminum and for
aluminum covered by carpet (for the case where another passenger space is above the
main deck). The aluminum overhead was found to reach a maximum temperature
of 55?C.
In order to determine the spread of fire to adjacent seats a series of FDS
simulations were run with adjacent seats instrumented within the model to deter-
mine the heat flux to the adjacent seats. If the heat flux seen by the adjacent seat
was greater than the the critical heat flux a theoretical time to ignition was calcu-
lated. The simulation was then run again with the ignition of this adjacent seats
programmed into the model, and the next adjacent seat instrumented for heat flux.
This process continued until the heat flux was not above the critical heat flux for the
29
Table 2.4: Comparison of USCG seat cushion data from testing per-
formed in 1998 and 2012.
Property 1998 Fire Test 2012 Fire Test
Heat of Combustion 21000 kJ/kg 17300 kJ/kg
Combustible Weight 1.0 kg 1.6 kg
Foam Density 45 kg/m3 38 kg/m3
Heat Released Per Seat 21000 kJ 27680 kJ
next adjacent seat. Using this method, a fire starting in a single seat was assumed
to spread to a maximum of 10 seats based on the limitation that the longest row of
seats will be 5. Two rows of 5 that are back to back are therefore assumed to ignite
in the fire scenario while rows facing each other or facing the same direction would
not spread as long as certain spacing is maintained. The study assumed that a fire
starting in a single seat will then spread only to a maximum of 10 seats. Based on
the study, performance guidelines were created to direct builders and operators.
These performance guidelines were published in Marine Safety Center Tech-
nical Note (MTN) NO. 01-13. The transient fire load is limited to 2.5 kg/m2 (0.5
lb/ft2) in order to prevent escape path obstruction and prevent the spread of the
fire to other seats. The construction and outfitting fire load is limited to 5.0 kg/m2
(1.0 lb/ft2). The gaps between rows of seats is also strictly set at a minimum of
30
76.2 cm (30 inches) for rows facing the same direction, and 45.7 cm (18 inches) for
rows facing one another. No more than 5 seats can be placed in a row and no more
than 10 seats can be arranged in an back-to-back formation. The total combustible
weight of each seat must not exceed 1.75 kg (3.86 lbs). In addition Vessels that
follow these guidelines are currently allowed to maintain their configuration without
adding structural fire protection insulation or completing a costly full engineering
analysis.
2.2.2 Swedish Baggage Testing
In 2010 a Swedish study was completed on the carried fire load of mass tran-
sit vehicles in an underground railroad system [3]. The increased need for safety
in transport systems, the general lack of available data on burning bags, and the
dangers of underground fires motivated the study.
Several rail fires were reviewed from catastrophic events in Sweden, Azerbaijan,
and South Korea. A study of these fires showed that passenger baggage was a factor
in the size of the fire and tenability conditions in the rail coaches and the tunnel.
The pictures in Figure 2.4 from the Baku metro fire in Azerbaijan clearly showed
a significant amount of passenger baggage and effects left in the unburned coaches
after evacuation [3]. The actual weight and composition of the baggage left behind
was not measured, but the amount shown in the picture clearly indicates that these
transient loads do have some effect on a potential fire.
Following a review of recent railroad fires throughout the world, the author
31
Figure 2.4: Picture of unburned rail-cars that were abandoned during
the Baku metro fire [3].
conducted a survey of metro and commuter trains in Sweden. The bags carried by
passengers at different times of the day and times of the week were recorded along
with the weight and general contents of the bag. A total of 622 bags were examined
during the study. The average weight of a bag on the commuter train was 4.4 kg
during weekdays and 4.9 kg during weekends. The average weight of a bag on the
metro was 3.5 kg during weekdays and 4.5 kg during weekends. On the commuter
train 87% of the people carried bags, while on the metro 82% carried bags.
Bags studied included: laptop, sports, tourist, school-university, school-high
school, handbag, suitcase, cabin bag, shopping bag, rucksack, pram (baby stroller),
trolley, and paper shopping bags. Representative bags were created and burned in
the laboratory to find the heat release rates. The ignition source for the test was a
32
pilot flame of 25 kW for 90 seconds.
The testing showed that the carried fire load in a metro train could be as much
as 50% of the fire load of the train itself. In addition, prams alone could possibly
be enough to cause flashover inside the train.
The weights of each of the bags, along with a weight breakdown for components
was recorded both before and after the test to determine what burned. The heat
release rate from each bag was measured during the testing and varied from 30 kW
for the trolley bag to more than 800 kW for the pram. In addition the total energy
released for each type of bag was calculated and compared to the theoretical total
energy of the bag before and after that fire. The heats of combustion used were 26.1
kJ/g for electronics, 19.0 kJ/g for textiles, 17.0 kJ/g for paper, and 47.0 kJ/g for
plastics. The data from the test is summarized in Table 2.5.
Commuter and metro trains were considered in the study. The average weight
per bag was calculated for both of the trains. The percentage of passengers that
carried bags was also calculated. Based on this, the average weight of the baggage
per person can be calculated. The floor area of the train car was not listed in the
study, but based on the per-person weight and the number of seats in a ferry the
weight of the baggage can be calculated in total and divided by the total deck area
to get the fire load density. The weight of each bag was broken up into electronics,
metal, textile, paper, plastic, and wood with the weight of each of these recorded
before and after the burn test to determine the contribution to the heat released.
The total weights were also recorded for each type of bag, as well as the heat released.
The handbag and suitcase both had sharp heat-release rate peaks corresponding to
33
Table 2.5: Relevant test dat from Swedish Bag Tests [3]. 1
Bag Type Peak Heat Release Rate (kW) Total Energy Released (kJ)
Computer Bag 110 45.2
Sports Bag 80 33.1
Tourist Bag 120 15.3
University Bag 85 15.5
High School Bag 65 7.9
Hand Bag 190* 14.7
Suitcase 720* 123.2
Cabin Bag 160 97.0
Shopping Bag 55 8.5
Rucksack 285 144.6
Stroller 830 179.1
34
the explosion of a pressurized can of hairspray. The peak was momentary and can be
largely ignored. The pram, which is the European name for a stroller, has by far the
highest heat release rate, reaching over 800 kW within 5 minutes. The possibility
of parking prams must be carefully considered, especially whether or not the pram
could be parked in the corner.
1Peak heat-release rates due to explosion of pressurized can of hairspray are momentary and
have been excluded from line fit data.
35
Chapter 3: Carriage Rates
3.1 Allowed Carriage
The allowed carriage rate for vessels currently stands at 2.5 kg/m2 (0.5 lb/ft2)
as laid out in MTN No. 01-13 for aluminum passenger vessels type 5A spaces
[19]. This transient fire load requirement has very little real world meaning for a
vessel operator that would much prefer to know how large or heavy a bag may be
carried aboard by each person. It would be more straightforward for operators and
regulators alike if there existed a policy or regulation similar to the airline industry
in which the size and/or weight of the baggage is limited by each carrier [20, 21].
Given the small administrative size of some ferry operators it would be difficult
to obtain good standard practices across the industry without a consistent policy
for operators to follow. The current policy and regulator requirements require some
calculation to interpret. Since the weight allowed per person aboard the vessel would
vary based on the passenger loading (because the deck area remains the same for
the calculation, the number of passengers would dictate the weight) the operator
could in theory use the following equation to find the answer:
36
Weight per Person =
Main Deck Area
No. of Passengers Scheduled
? 2.5 kg/m2 (3.1)
However, this only works if the operator knows ahead of time how many people
will be present on the trip. If true, this would still leave the logistical problem of
informing passengers ahead of time of the weight allowed per person, and ensuring
that they knew it meant baggage would be stored separately or left behind.
The calculation above may be easy for an ocean excursion vessel, or an air-
line trip where booking is required well ahead of time, but for most ferry vessels
booking is not necessarily required in advance. Many operations allow tickets to be
purchased immediately preceding a trip, running on a schedule much like that of a
bus. This is a major part of the business plan and must be taken into consideration.
Thus, for these operators it is impossible to run a calculation ahead of time, and
the logical assumption from a conservative safety perspective, is to complete the
calculation assuming a full load of passengers. This equation looks exactly like the
previous equation with the number of scheduled passengers being replaced by the
total number of seats, which results in a full load assumption.
Weight per Person =
Main Deck Area
No. of Seats
? 2.5 kg/m2 (3.2)
The weight of baggage allowed for each person then depends upon the number
of seats and the deck area. These two figures can be combined into the seat density,
which is a measure of the quotient of these values. Given the required maximum
transient fire load of 2.5 kg/m2 (0.5 lb/ft2), and the seat density of the vessel, the
37
Figure 3.1: Allowed weight per passenger based on seat density and
using a conservative full load assumption. The red dashed lines bound
the normal range of seat densities found in currently operating ferry
vessels.
weight of baggage per person can be calculated for each vessel at the assumed full
load condition. Figure 3.1 shows the allowed weight per person based on varying
ferry seat densities. The normal range of seat densities shown between the red
dashed lines corresponds to a normal range of allowed weight per person of 1.4 to
3.8 kg (3.1 to 8.5 lbs).
This figure will be compared with the findings to see if vessels are currently
falling within the requirements by way of ferry company policies, or if these policies
are inadequate to the task of enforcing a somewhat abstract regulatory requirement.
38
The normal range of allowed weight per person as seen in Figure 3.1 is a realistic
value when considering the average weight of a bag, but is quite restrictive in limiting
every bag to a maximum of only 3.8 kg (8.5 lbs) or less depending on seat density.
3.2 Data Collection
In order to ascertain the current state of the carriage of baggage aboard alu-
minum ferry vessels, data was collected from these vessels. Although the weight
of baggage brought aboard per square meter of space was the ultimate desire, the
carriage rate, baggage type distribution, and weight of individual bags was of in-
terest as well. A total of three vessels were surveyed in the southern New England
area, and a total of 12 runs on these ferries was considered. The ferry vessels were
divided into two types, commuter and non-commuter ferries. Commuter ferries are
considered as usually short-run vessels which carry primarily passengers that are
utilizing the ferry as a means of traveling to and from their place of work. It is
suspected that these vessels may have a lower weight of baggage brought aboard
as passengers will typically not be carrying clothing, toiletries, or long-term items.
The non-commuter ferry vessels include all other ferry vessels including those that
take day trip passengers, weekenders, and even passengers that may be travelling
for extended vacation periods. The data was collected on a Thursday through Sun-
day during the summer (August 22,23, and 24); which is important to ensure that
the non-commuter ferries are taking a large number of vacationers and beach-goers
indicative of a worst case loading condition.
39
Two data sets were collected for each ferry vessel run. The first data set
was collected by obtaining data points at the boarding location of the ferry vessel.
Every bag that was brought aboard the vessel was classified as one of the following:
tote, purse, small duffel, carry-on suitcase, large duffel, large suitcase, backpack, or
briefcase as shown in Figure 3.2. In addition there was a data collection column for
no bag, which was included in order to ascertain the overall carriage rate. The second
data set was collected during the transit of the ferry by weighing and photographing
a sampling of baggage of each type aboard the vessel. Passengers were solicited at
random and were given the option of volunteering to have their baggage weighed
and photographed. The bags were once again classified by type in order to obtain
a weight distribution for each baggage classification.
Although data collection was completed on large suitcases and duffel bags,
both non-commuter ferry operators had company policies in place whereby these
larger bags were separated from the passenger and placed in special luggage storage
compartments. Although several data points were taken on the weight of these
larger baggage items, the carriage of these bags was eliminated from the carriage
rate and average weight data since the bags were not present in the compartment
of interest.
3.2.1 Commuter Ferry
A ferry that completes commuter runs from Hingmham, MA to Boston, MA
was studied for two early morning commuter runs at 7:15am and 8:45am as well as
40
a 8:00am return trip between the runs. At 7:15am there were a total of 193 people
surveyed boarding the vessel, 81.6% carried bags of some kind. Of those carrying
bags 43.6% carried briefcases, 28.8% carried purses, 22.1% carried backpacks, and
the remaining 5.5% carried some form of duffel bag or suitcase. At 8:45am 148
people were surveyed, with 81.1% carrying bags of some kind. Of those carrying
bags 37.5% carried briefcases, 35.0% carried purses, 24.2% carried backpacks, and
the remaining 3.3% carried some form of duffel bag or suitcase. This data follows
the expected trend for a commuter vessel as backpacks and briefcases are preferred
methods of carrying work requisites, and purses are commonly carried on a daily
basis.
The average weights of a backpack and briefcase brought aboard the vessel
were 4.8 kg (10.5 lbs) and 4.8 kg (10.6 lbs) respectively. This average value exceeds
the normal limiting value of 3.8 kg (8.5 lbs), with some briefcases and backpacks
weighing in at more than 6.8 kg (15 lbs). The average weight of a purse for the
commuter run was only 2.5 kg (5.4 lbs), which falls within the range of allowed
weight, but only if the seat density remains at or below 1.0 seats per square meter.
Additional data collected from the survey is presented in Appendix A.
3.2.2 Non-Commuter Ferry
Two different non-commuter ferry vessels were studied; one that takes pas-
sengers from Point Judith, RI to Block Island, RI, and one that takes passengers
from Hyannis, MA to Nantucket Island, MA. These two islands are common sum-
41
mer destinations for day-trippers, beach-goers, and passengers going on vacation for
several days up to a few weeks. As a result, the luggage carried by the passengers,
particularly during the summer months, can be quite significant and the ferries are
usually filled to capacity on the weekend days. Data was collected from Point Ju-
dith, RI to Block Island, RI on a Friday from midday through early afternoon, and
the data from Hyannis, MA to Nantucket, MA was taken on a Saturday from late
morning through mid-afternoon. The date, day of the week, and time are critical to
ensuring that the data collected is indicative of the greatest passenger and baggage
loading conditions seen by the ferry vessels throughout the year. Non-commuter
data was collected for more than 1000 passengers with a carriage rate of 88.5% over
all ferry runs. Of the passengers carrying baggage 32.2% carried a tote bag, 30.4%
carried a backpack, 18.1% carried a purse, and the remaining 19.3% carried small
suitcases, duffel bags, coolers, or briefcases. These results are very similar to the
commuter ferry results in terms of type percentage, with the exception that the car-
riage rate of briefcases in the commuter ferry has been replaced by the tote bag for
the non-commuter run. Since more passengers are using this vessel for overnight and
long-term travel the percentage of suitcases and duffel bags is significantly higher
for the non-commuter ferry as expected.
The average weight of the tote bag and backpack were 4.2 kg (9.3 lbs) and
5.0 kg (11.1 lbs) respectively. As with many bags carried aboard the commuter
ferry, these values exceed the normal limiting value of 3.8 kg (8.5 lbs) from Figure
3.1. The average weight of a purse for the non-commuter run was 2.5 kg (5.6 lbs),
which is consistent with data taken from the commuter ferry vessel. This weight
42
falls within the allowed weight range of baggage per person assuming full load, but
only if the seat density remains at or below 1.0 seats per square meter.
3.3 Results
The focus of the baggage data collection was to ascertain the current baggage
carriage aboard ferry vessels and determine if this level of carriage falls within the
current regulatory requirement of 2.5 kg/m2 for transient fire load. The average
weight per bag carried aboard each type of ferry was multiplied by the carriage rate
to determine an average baggage weight per person boarding the ferry. The average
baggage weight for the commuter ferry was 3.4 kg (7.4 lbs); giving an average weight
per person of 2.8 kg (6.2 lbs). The average baggage weight for the non-commuter
ferry was 4.2 kg (9.2 lbs); giving an average weight per person of 3.7 kg (8.1 lbs).
These numbers are in close agreement with the average weight per person calculated
by Kumm [3] for carriage on trains. The average weight per person on a metro train
on a weekday, comparable to the commuter ferry on a weekday, was 2.9 kg per person
as compared to 2.8 kg per person for the ferry. The average weight per person on a
commuter train on a weekend, comparable to the non-commuter ferry on a weekend,
was 4.3 kg per person as compared to 3.7 kg per person for the non-commuter ferry.
The vast majority of baggage carried, 79%, weighed less than 6.8 kg (15 lbs)
with approximately half (54%) weighing less than 4.5 kg (10 lbs) as seen in Figure
3.3. The frequency of the baggage weights measured shows that the average value
calculated is consistent with the greatest frequency of baggage weight carried. The
43
Figure 3.2: Representative samples of baggage that were weighed on
commuter and non-commuter ferries.
44
Figure 3.3: Histogram showing the weight of baggage and cumulative
percentage carried aboard ferry vessels.
45
largest group percentage of 35% for the 5 to 10 lb range, easily encompasses both
the average values calculated. The weight distributions of each bag type, along with
additional survey results are shown in Figures A.1 through A.13.
Next the average weight per person must be compared with the seat density
of the vessels studied to determine if the requirements of the regulations are being
met. The calculated weight along with seat density information obtained from the
builders of the vessels will be inserted into Equation 3.3 to calculate the currently
carried weight per square meter.
Weight per Person? Seat Density = Weight per Square Meter (3.3)
Table 3.1 shows the results of the calculations for all three of the ferry vessels
surveyed. All three vessels, as loaded at the time of the data collection, exceed the
transient fire load requirement of 2.5 kg/m2 (0.5 lb/ft2). Using the transient fire
load requirement and the average weight per person taken from data collection, the
cutoff seat density can be calculated to get an intimation of how much of the current
aluminum ferry fleet is exceeding the requirements on a busy day with a full load of
passengers.
Given the commuter ferry average weight per person of 2.8 kg, the seat density
would have to be equal to or less than 0.89 seats/m2 in order to meet the regulatory
requirement. Data on seat densities for vessels with Type 5A spaces indicates that
only 28.8% of currently operating vessels have a seat density of less than or equal
to 0.89 seats/m2. Given the non-commuter ferry average weight per person of 3.7
46
Table 3.1: Seat density and transient fire load for three ferry vessels from
which carriage data was collected.
Ferry Vessel Seat Density (Seats/m2) Transient Fire Load (kg/m2)
Commuter Ferry 0.99 2.77
Non-Commuter Ferry 1 0.79 3.08
Non-Commuter Ferry 2 1.07 4.17
kg, the seat density would have to be equal to or less than 0.68 seats/m2 in order
to meet the regulatory requirement. Only 3.4% of vessels meet this seat density
requirement, indicating that almost any ferry vessel operating as a non-commuter
ferry could be exceeding the requirements on its busiest days. In order to ensure that
all vessels and seat densities were operating within the requirements, the allowed
baggage weight per person would have to be limited to 3.1 lbs, an unreasonable
number when considering that of the baggage surveyed 95.7% exceeded 3.1 lbs.
Excluding more than 95% of currently carried bags would be excessive, particularly
during off-peak times.
The carriage rate of the different bags will be used to weight a heat-release
rate curve using data from Kumm, with an adjustment made for the average weight
difference. This will allow simulations to be conducted to test if the current level of
carriage will cause sufficient heat transfer to compromise the structural integrity of
the aluminum superstructure.
47
Chapter 4: FDS Model
4.1 Verifying the Mechanism
In order to create results that are useful to policy making bodies within the
Coast Guard, the mechanism used to create and track the progress of the fire in
the simulated space must closely match the mechanism used in previous research by
Shriner [8]. The overall physical dimensions of the space are known, along with the
basic type layout of the seat banks. Measurements for the seat area will be used
to create the physical model of the seats, but since FDS will not be used to model
the spread (hand calculations will be used) the precise properties of the seats are
relatively unimportant. The seats will be modeled as non-combustible so that spread
does not occur within the simulation. Heat flux gauges will be used to measure the
incident heat flux at the seat cushions adjacent to the seat on fire, and this flux will
be used to predict the time of spread, which will be programmed into a subsequent
FDS simulation. In this way each fire scenario will be run in a series of steps as the
spread is calculated and then modeled into the next simulation step in the scenario.
The two most important parameters to be matched between this model and
the model used by Shriner are the specification of the fire and the placement of the
heat flux gauge. The fire is specified using the HRRPUA (Heat Release Rate per
48
Unit Area) command in FDS with a ramp function used to control the growth and
decay of the fire for each individual seat [15]. The ramp function within FDS allows
the user to specify independent and dependent variables so that quantities such as
the heat-release rate can be ramped up according to a set of simple linear equations.
Full scale testing from Shriner [8] provides a ramp function for the heat-release rate
for middle ignition and side ignition of the seats, as shown in Figure 4.1. The area
of the seat is known, along with the size of the fire and the progression of the heat
release rate, so that the fire is fully specified and matched with the work of Shriner.
The next task is to match the heat flux gauge measurements that were recorded
during the simulation and used to calculate the spread of the fire between adjacent
seats. The exact mechanism used by Shriner for measuring the heat flux in FDS
is not known. The location of the heat flux measurement point was specified as 7
cm from the fire and will be assumed to be a straight horizontal measurement from
seat cushion to seat cushion. The vertical placement of the point was not specified,
nor was the orientation, except to say that the gauge faced the fire. A total of 3
vertical placement and 3 orientations were used to record the heat flux at the next
adjacent seat to the seat on fire. The mechanism that gave a result most closely
matching the work of Shriner was the radiative heat flux gas quantity placed at a
90 degree orientation to the seat on fire at a distance of 7 cm and a height of 0.5
meters, which corresponds to the height of the top of the modeled seat cushion. The
quantity called ?radiative heat flux gas? within FDS acts to place a radiometer at
the specified location and integrates the incoming radiative flux over a solid angle
of 2pi, centered around the specified orientation vector [15].
49
Figure 4.1: Experimental measurements and ramp functions for middle
and side ignition of seats given by Shriner [8].
50
Although Shriner?s data on heat flux gauge measurements was unavailable for
direct comparison, the ignition curve and calculated ignition time for adjacent seats
provided a method of comparison. The specified ignition curve from Shriner, showed
a relationship between the inverse of the square root of the ignition time, tig and the
natural log of the heat flux, HF and was used to calculate an ignition time for the
adjacent seats of 200 seconds for Shriner?s work [8]. By eliminating heat flux values
below the critical heat flux of 10 kW/m2 and taking an average heat flux value over
time for use in the ignition equation, the ignition time can be recalculated using
heat flux gauge measurements from current simulations with the radiative heat flux
gas quantity. The calculated ignition time for adjacent seats from the simulation
data is 190 seconds, giving an error of 5% for the ignition time.
t?1/2ig = 0.21 lnHF ? 0.46 (4.1)
One of the reasons for this error is the difference in the FDS grid resolution
between the present work and the work of Shriner. Additional time available for the
present work allowed for a finer grid resolution, which in turn creates small differ-
ences in data as the calculations are refined and improved. The error is small (4.5%),
and because the calculated ignition time is shorter than that shown by Shriner, this
calculation will give a more conservative result, which is often desired when making
life safety decisions. Additionally, the increased grid resolution suggests a further
degree of accuracy in the current measurements.
51
4.2 Grid Resolution Study
An important aspect of setting up a CFD simulation is to ensure that the sim-
ulation has the proper grid resolution. The grid resolution is a measure of how well
resolved the problem is; meaning how many grid cells are present for a characteristic
length scale. The more grid cells that are present for a given length scale, the finer
the mesh and the greater the resolution of the problem. Although greater resolution
means theoretically greater accuracy of the problem, the computational costs and
potential numerical errors increase rapidly with increased grid resolution. Doubling
the grid resolution, for instance, will result in roughly 16 times the computational
time, so increases in resolution must be carefully considered and sensitivity must be
studied to ensure adequate resolution without wasted computational cost. The FDS
User Guide suggests using Equation 4.2 to calculate the characteristic diameter of
the fire. This characteristic diameter is used as the characteristic length scale for
the problem and is divided by the cell size in order to obtain a grid resolution value.
D? =
(
Q?
??cpT?
?
g
) 2
5
(4.2)
D* = characteristic fire diameter
Q? = heat-release rate (kW)
?? = density of air (kg/m3)
cp = specific heat capacity of air (J/kgK)
T? = ambient temperature (K)
g = acceleration due to gravity (9.81 m/s2)
52
The characteristic fire diameter was calculated using 700 kW as the peak value,
as this is the peak sustained heat-release rate from the burning of a stroller, the
highest single item heat flux. Using the properties of air and the peak heat-release
of 700 kW yields a characteristic fire diameter of 0.831. Using a course grid resolution
value of 4 would yield a mesh of 135?60?20 with a total of 162,000 cells; while a
finer grid resolution value of 10 would yield a mesh of 324?150?48 with a total of
2,332,800 cells. A grid resolution value of 10 is preferable to ensure better accuracy,
but the resolution must be carefully checked to compare the level of accuracy increase
and the increase in computational cost. Given that supercomputers and multi-
computer banks are not available for this work, the computational cost will be a
significant limiting factor in deciding on the final grid resolution.
For the characteristic fire diameter of 0.831 the grid resolution value was tested
at 8.3, 10, and 11.5 to check for grid convergence. Since the quantity of interest in
this testing is the ceiling temperature, the average ceiling temperature was compared
between the three different mesh resolution values. This comparison can be seen
in Figure 4.2. The grid resolution value of 10.0 differed from the grid resolution
value of 8.3 by 2.6% when considering the peak average ceiling temperature. The
grid resolution value of 11.5 differed from the grid resolution value of 10.0 by 0.5%
when considering the peak average ceiling temperature. Although there is some
advantage in increasing from a grid resolution value of 8.3 up to 10.0, the advantage
when raising the value above 10.0 decreases rapidly. A greater increase above 10.0
to a value of 15 or higher may have a very slight advantage, but the computational
cost of running a simulation at this resolution is not possible due to time constraints.
53
Figure 4.2: Grid resolution comparison showing the average ceiling tem-
perature across the passenger compartment for three different grid res-
olution values. The ceiling temperature (the important parameter for
this study) converges at a grid resolution value of 10.
The grid resolution value of 10 will be used yielding a final mesh of 324?150?48
with a total of 2,332,800 cells.
4.3 Creating an Average Bag
One goal of this research is to determine a single number value that depicts
the allowed carriage of bags on a ferry. Even with only 6 bag types and a seat bank
that includes 10 seats, the number of permutation for baggage arrangement is in
the millions and far greater than is possible to test completely for this work. It is
54
desired therefore to create an average scenario in terms of baggage placement. In
order to do this there must be a single average bag with a specified heat release
rate curve that can be placed in conjunction with the heat release rate curve of the
seat. In order to create this average bag the carriage percentage for ferry vessels will
be used in conjunction with the average weight of a bag to determine the average
heat-release rate.
Baggage heat-release rate data was taken from the Kumm study [3] and used to
create linear heat-release rate curves that specify the heat-release rate using between
four and eight lines described by a series of time and heat-release rate coordinates.
The area under these created curves was compared with the total heat released
during testing of the baggage to ensure that the area under the curve matched the
area under that data points given. Since each curve was specified using different
points the time coordinates used to describe each curve were also different. Linear
interpolation was used to ensure that for each unique time coordinate present for
any bag, all bags had an interpolated data point for heat-release rate at that time.
The heat-release rate approximation curves for all baggage are shown in Figure B.1
through B.6. The result was a heat-release rate value for every baggage type at each
possible time coordinate. These heat-release rate values were then weighted based
on the carriage rate of the particular bag type to get a single heat-release rate value
for each time coordinate that describes the average heat released. The resulting
average heat-release rate curve can be seen in Figure 4.3.
Now a correction must be made to the data based on the differences in mea-
sured weight of baggage between the Kumm study and the data taken from ferry
55
Figure 4.3: Weighted heat release rate curve for an average bag carried
aboard a ferry.
56
vessels. The average weight of each baggage type is taken from Kumm and weighted
based on the carriage rate from vessel surveys. The summation of these values gives
the average weight based on the baggage weight given from the Kumm data of 2.4
kg (5.3 lbs). This compares to an average baggage weight of 3.7 kg (8.1 lbs) from
vessel surveys. The heat-release rate at each time coordinate is then multiplied by
the quotient of the average weight from vessel surveys and the average weight based
on the Kumm study. This gives a curve that represents the heat-release rate curve
of an average bag brought aboard a ferry vessel. This heat-release rate curve can
be used alone to simulate a bag placed on the deck of a ferry or it can be added to
the heat-release rate curve for a burning seat to create the scenario of baggage left
behind on top of seat cushions.
4.4 Scenarios
A total of three scenarios were considered for this research. The first two
configurations are scenarios which use an average bag heat-release rate combined
with a particularly concerning baggage placement location to present dangerous,
but not necessarily worst possible case scenarios. Due to the restricted entrainment
into the fire plume caused by the corner configuration, a corner fire can have a
significantly greater impact on overhead heating than a fire that occurs in the open
due to the increased flame height and potential direct impingement on the overhead.
To test this dangerous placement scenario, a stroller was parked in the corner, as
had been seen during vessel surveys (seen in Appendix B), to determine if corner
57
storage of baggage has the potential to compromise the structural integrity of the
aluminum. Next the normally clear escape aisle between seat banks was occupied
by a carry-on type bag to determine if the possible ignition of multiple seat banks,
filled with average bags, would create a large enough fire to raise the aluminum
overhead temperature above prescribed safe limits.
Finally a more common scenario was created that involved the burning of
one seat bank consisting of ten total seats arranged in two back to back rows of five
seats each. This scenario represents the current seat bank size limitation imposed on
builders who wish to use the Coast Guard exemptions from structural fire protection
insulation. This scenario will include the addition of the average bag on top of each
seat within the seat bank. This scenario will be tested to determine the extent of
heating of the aluminum overhead, and to ascertain whether the addition of bags
on the seats creates a fire big enough to ignite additional seats banks around the
initial fire.
4.4.1 Stroller in the Corner
It was observed during ferry vessel surveys that strollers were present on nearly
every non-commuter ferry run. The placement of these stroller was concerning as in
some cases they were placed in escape aisles, against walls, or even in compartment
corners. The study by Kumm gave a peak heat-release rate of a stroller (called
a Pram in that study) of 830 kW, with a total energy released of 179.1 MJ [3].
This energy content is greater than any of the other single items of baggage and
58
is therefor a significant concern. Although a stroller parked within an escape aisle
could potentially spread the fire to several adjacent seat banks, this scenario is seen
as not occurring frequently as most strollers are stored elsewhere and due to the
large size of a stroller relative to a bag it could not be stepped over during escape
and would have to be pushed or otherwise moved out of the aisle during vessel
evacuation. The dangerous common scenario is for a stroller to be parked in the
corner of the space as seen in Figure 4.4. The stroller was set up as a simple box
in FDS with a prescribed heat-release rate according to an accompanying ramp
function, done to maintain consistency between the method of modeling the seats
and the baggage. The stroller box was given dimensions of 1.0 meters (39.4 inches)
long by 0.5 meters (19.7 inches) wide by 1.0 meters (39.4 inches) high, and was placed
against each wall of the corner to simulate being parked in the corner. Temperature
measurement devices were placed on both of the corner walls as well as the ceiling
above the burning stroller.
4.4.2 Burning of Two Seat Banks
Although bags are normally restricted from being placed in escape aisles, it is
possible that baggage could be placed there or left there during evacuation, leading
to the potential ignition of seat banks on either side of the escape aisle. To test
this scenario a carry-on bag was modeled in FDS as a 0.56 meter (22 inches) by
0.36 meter (14 inches) carry-on type bag placed on the deck at the exact midpoint
between two seat banks as shown in Figure 4.5. A carry-on type bag was selected
59
Figure 4.4: FDS setup for burning of the stroller in the corner.
60
because it has the highest heat-release rate of any bag (stroller excluded) that may
be carried into a passenger compartment. The width of the aisle was made to be
0.76 meters (30 inches), which is the maximum aisle width allowed by Coast Guard
regulations. The maximum possible aisle width was used for this simulation because
if this aisle width allows the ignition of adjacent seat banks, a smaller aisle width will
also certainly cause ignition. The adjacent seat banks each consist of ten total seats
with each bank being arranged in two rows of five seats placed back to back. This
creates the tightest concentration of seats possible within the current requirements.
Radiative heat flux gas gauges were placed on each of the four seats closest
to the burning carry-on bag, with the placement being as near center line of the
seat bank as possible. The heat flux gauges can be seen in Figure 4.5 on the left
hand side as green dots. If the heat flux is recorded at any time as exceeding the
critical heat flux value of 10 kW/m2 than the ignition equation will be utilized to
determine if the seats ignite. If the seats ignite, the simulation will be reprogrammed
with burning seats and the heat flux gauges would be moved to the next adjacent
seat. This process will continue until all seats are burning, until the fire no longer
causes ignition of the next seat, or until critical temperatures are reached for the
aluminum structure. The average bag heat-release rate will be added to any seats
that are determined to ignite so that this scenario will consider a fully loaded ferry
vessel.
61
Figure 4.5: FDS setup for burning of two seat banks. The burning carry-
on bag is shown by the orange box in the middle, with the seat cushions
shown in blue. Heat flux gauges for the left hand seats are visible as two
green dots, which are oriented to face the fire.
4.4.3 Burning of a Single Seat Bank
The final scenario to be tested is the burning of a single bank of ten seats with
the addition of an average bag on each seat. The ten seat bank consists of two rows
of five seats each, placed back to back. This is the largest single seat bank allowed
by the current exemption policy requirement given in MTN 01-13. A ten seat bank
of the same size was tested by Shriner without any baggage or additional loading
and found to raise the ceiling temperature to 55?C. The addition of the bags to
the seats will determine what temperature the ceiling will reach with the additional
heat released from an average bag. This will aid in determining if the carriage
of baggage at its current level, as measured during vessel surveys, is sufficient to
raise the ceiling temperature over any square meter to a critical value of 200?C. In
addition heat flux gauges were placed on each of the seats across aisles and across
62
Figure 4.6: FDS setup for burning of a single seat bank. The initial seats
set on fire are indicated as orange seats, with heat flux gauges shown as
green dots.
knee gaps to determine if any adjacent seat banks will ignite now that the seat bank
fire is larger with the addition of bags. Heat flux gauges were placed on the closest
edge of each seat oriented to face the fire and located as shown in Figure 4.6.
63
Chapter 5: Results & Analysis
5.1 Simulation Results
5.1.1 Stroller in the Corner
The simulation of the stroller placed in the corner of the space reached its peak
heat release rate 120 seconds into the simulation and maintained this heat release
rate until 240 seconds, at which point the heat release rate began to decline sharply.
The peak ceiling temperature closest to the corner was 821?C, a value sharply in
excess of the 555?C melting temperature of the 6082 Aluminum alloy commonly
used in ship superstructures [1].
Figure 5.1 shows the ceiling temperature measured above the burning stroller
in the corner of the space. The melting point of the material is also shown, along
with the temperature at which the aluminum has lost approximately 50% of its
structural integrity [22]. The aluminum ceiling maintains a temperature above the
melting temperature for a period of 156 seconds and maintains a temperature above
the 50% loss of structural integrity for a period of 367 seconds. According to MTN
01-13 no single point of the aluminum deck shall reach 400?C and the temperature
over any square meter must no reach 200?C. Both of these limitations are exceeded
64
Figure 5.1: Corner ceiling temperature above burning stroller.
65
in the ceiling of the passenger compartment in which the stroller fire takes place in
a corner.
In examining the results of the stroller fire it is important to consider the size
and type of the stroller that is modeled in the space. The study that provided
the heat-release rate data used to model the stroller fire was conducted in Sweden,
meaning that the type of stroller may be different than those used in the United
States. The weight of the stroller used for the full-scale testing in Sweden was 15.1
kg, while strollers in the United States are often as lightweight as 5.0 kg. Although
strollers used in the United States can weigh as much as 20.0 kg, using data from
a 15.1 kg stroller means that the simulation conducted leans towards a worst case
scenario with a heavy stroller. The stroller used in the Swedish study was also
constructed with a metal frame weighing 5.4 kg; which contributes nothing to the
heat-release rate, while a plastic frame frequently used in today?s strollers in the
United States may contribute significantly to the heat-release rate of the stroller.
In addition to the stroller itself, there could be notable weight increases depending
on the blankets, pillows, and diaper content of the stroller or the bags that could
be attached or stored underneath in stroller storage bins. Observations of a stroller,
filled with goods adjacent to a pile of newspapers and waste basket in the corner
of a surveyed vessels raises concerns that the modeled heat-release rate could be
reached in a real passenger compartment.
66
5.1.2 Burning Bag In Aisle
For the second scenario, the carry-on bag placed in the center of the escape
aisle caused the nearest seats to exceed the critical heat flux of 10 kW/m2 at 234
seconds. The heat flux continued to increase and at 306 seconds the heat release
rate per unit area reached 19.7 kW/m2, creating an average over 72 seconds of 15.8
kW/m2. Using Equation 4.1 from Shriner, an average heat flux of 15.8 kW/m2 will
cause ignition in 69.9 seconds, indicating that all four seats adjacent to the carry-on
bag will ignite from the side at 306 seconds.
Following this same process, the next four seats ignite at 515 seconds, the
following four seats at 725 seconds, and the last four seats in the two seat bank
ignite at 1148 seconds. Figure 5.2 shows the heat-release rate from the fire with the
ignition point of each subsequent ?set? of four seats indicated. Although there are
several peaks, the highest heat-release rate peak of 1255.5 kW occurs 966 seconds
into the simulation. The peak is reached at this time because it corresponds to the
time at which the first four burning seats decrease to a heat-release rate of only 60.1
kW/m2 from their peak at over 200 kW/m2. At this point the first set of four seats
is effectively burned out, and the second set of four seats has dropped below half of
its peak heat-release rate. The average temperature of the aluminum overhead in
the area of the fire at this time is 161?C, and continues to increase.
The two criteria for the aluminum ceiling temperature are that the aluminum
not reach a temperature of 400?C at any point, and that the average over any
square meter not reach a temperature of 200?C. These values are taken from the
67
Figure 5.2: Heat-release rate from a simulation of burning of two seat banks.
68
Figure 5.3: Ceiling temperature profile above two burning seat banks at
1434 seconds after ignition of the carry-on bag.
policy requirements published in MTN 01-13. The maximum single point ceiling
temperature reached during the simulation is 363?C, a highly elevated temperature,
but still within the requirements.
The average temperature over a square meter was not considered in the center
above the carry-on bag, but rather centered above the outer four seats to burn on
the lower end in Figure 5.3. Since the carry-on baggage burns out first, the ceiling
directly above it does not reach the highest temperature. Also the relatively low
density of fire load in the area of the bag, particularly given the openness of the
aisle and the low height of the fire due to the bag being on the floor, means that
69
Figure 5.4: Average temperature over the square meter of ceiling above
the lower four burning seats.
70
the fire does not heat the ceiling to critical temperatures at that point. The ceiling
above the lower and upper four seats at either end of the seat banks is essentially
preheated by the carry-on and first six seats, after which the temperature peaks
when the last seats are ignited. This shapr increase can be seen in Figure 5.4, which
shows the average temperature of the ceiling above the lower four seats of the two
seat banks. The average temperature of the square meter above these seats reaches
a maximum value of 257?C at a time of 1420 seconds, just under 24 minutes into
the simulated fire. This value clearly exceeds the maximum of 200?C over a square
meter, and also exceeds the temperature at which there is a 50% loss of structural
integrity for 6082 Aluminum.
5.1.3 Single Burning Seat Bank
The single seat bank simulation was conducted assuming that a burning bag
or other heat source tucked underneath the seat causes middle ignition of the two
back to back seats at the center of the 10 seat bank. The heat flux measured at the
adjacent seats reached 10 kW/m2, a critical value, 116 seconds into the simulation
and had an average heat flux of 16.1 kW/m2 over the next 65 seconds leading to
ignition of the next seats at 181 seconds. With a total of six seats now burning,
the remaining four adjacent seats in the current bank ignite at 385 seconds. The
ignition times were calculated using Equation 4.1 as before.
Even with all 10 seats and their associated baggage burning, the ceiling tem-
perature above the single seat bank does not reach critical levels; therefore, the
71
Figure 5.5: Heat release rate from a single bank of burning seats with baggage.
72
Figure 5.6: Average ceiling temperature over one square meter above
burning single seat bank.
potential for spread across the knee gap on either side must now be considered.
Heat flux gauges were placed on the leading edge at the middle of seats across the
knee gap on either side of the burning seat bank. The middle seats across the knee
gap receive critical heat flux starting at 348 seconds, and ignite at 419 seconds. The
next four seats across the knee gap ignite a mere 30 seconds later at 449 seconds into
the simulation. The quick succession of ignitions is seen in Figure 5.5 and causes an
ensuing spike in the heat-release rate and ceiling temperature of the compartment.
The two criteria for the aluminum ceiling temperature are again that the
aluminum not reach a temperature of 400?C at any point and that the average over
73
any square meter not reach a temperature of 200?C. The maximum single point
ceiling temperature reached during the simulation is 291?C at 725 seconds, safely
below the maximum allowed temperature of 400?C. The average temperature over
any square meter was considered in the center directly above the middle seats in
the single bank of seats. The average temperature of the square meter above these
seat, as shown in Figure 5.6, reaches a maximum value of 253?C at a time of 700
seconds, only 11.7 minutes into the simulated fire. This critical temperature is
reached quickly for the single seat bank simulation because of the quick succession
of seat ignitions across the knee gap between seat banks.
5.2 Analysis of Results
5.2.1 Evacuation Carriage
An important consideration in determining the contribution of baggage to a
fire in a passenger compartment is how many bags are removed from the space by
passengers during evacuation. The size and openness of a ferry vessel passenger
compartment, in comparison to an airplane, provides little motivation to study the
carriage of baggage during evacuation; and it has not been exhaustively studied.
In the commercial airline industry the carriage of baggage during evacuation is of
significant concern. Flight attendants are trained to instruct passengers to leave
behind baggage during evacuations and in some cases are trained to take baggage
from passengers as they are leaving to ensure that bags do not cause evacuation
delays. In addition, airplane evacuation tests include piles of carry on luggage in
74
the aisle to simulate the delays caused by baggage. Although the present research is
not concerned specifically with evacuation time, there is an undeniable connection
between the carriage of baggage during evacuation and the transient fire load in the
passenger compartment that is the subject of this research.
The current simulations were completed with the assumption that all of the
baggage brought aboard the ferry vessel remains in place in the passenger compart-
ment during the evacuation of the passengers. This would be the worst possible
situation in terms of the transient fire load, and thereby gives a conservative esti-
mate which is often desired when making life safety decisions. However, if a large
percentage of people take baggage with them during evacuation, this may have a
potentially huge impact on the size of the fire in the compartment and the potential
for spread between seat banks.
An airline survey conducted by the National Transportation Safety Board
(NTSB) in 2000 indicated that 46.8% of respondents who had been involved in an
airplane evacuation attempted to retrieve baggage prior to evacuating the airplane
[23]. In the airplane environment it is harder to carry a bag because of tight aisles,
and retrieval can be difficult and time consuming due to overhead compartments;
thus there are a number of factors to discourage people from taking baggage with
them during a plane evacuation. These same discouragements are not present on
a ferry vessel, and the carriage rate during evacuation could potentially be even
greater. If baggage is stored at people?s feet or in the seat next to them it is second
nature to retrieve the bag while getting up to begin evacuation. Passengers that
retrieved baggage during an airline evacuation reported that they needed to retrieve
75
bags because of medicines, job items, keys, wallets, and credit cards. These same
motivations will be present during a ferry evacuation and will provide incentive for
people to take bags with them during evacuation.
Ship evacuation studies typically focus on larger vessel such as cruise ships
and are more concerned with response and evacuation time rather than carriage of
baggage or purses, so very little data exists on carriage for ferry vessels [24?26].
However, given the natural desire of people to take bags with them, particularly
if they have money, wallets, medicine, or important papers in the bags, it can
be surmised that at least some people will try to take baggage with them during
evacuation. That means that although the exact amount cannot be pinpointed
without further research, it is known that some percentage of people will attempt
to take baggage with them during evacuation. Given that airline studies indicate
almost half of people carry bags with them, means that a factor of safety as high
as two may be built in to the simulations given the assumption of all bags remain
within the space.
5.2.2 Extent of Simulations
The single seat bank burning has a peak average temperature over a square
meter of 253?C, a temperature equal to that created by the burning of two seat
banks. In addition the heat-release rate for the single bank of seats is significantly
greater than that of the two seats banks. The reason for this is the methodology
used to calculate the spread of the fire along the seat banks. The procedure used was
76
to first consider the burning of the seat bank or banks in question, determine if the
ceiling temperature reached critical values, and then consider ignition of seats across
the knee gap only if the initial seat bank(s) did not reach critical temperatures.
The single seat bank fire did not reach critical temperatures and so the spread
across the knee gap was considered. This knee gap spread led to the ignition of
six additional seats within 30 seconds of simulation time, and thus created a quick
peak temperature that easily exceeded critical values. The two burning seat banks
on their own exceeded the critical value of 200?C and thus the knee gap spread
was not considered. Were this type of spread considered the heat-release rate and
temperature of the two seat banks should easily exceed that of the single seat banks.
Since the goal of the research was to consider the pass/fail criteria of the critical
values, and due to time constraints caused by long simulation times, the full extent
of the fires was not considered past the critical average ceiling temperature. This
means that the burning of two seat banks, with knee gap fire spread considered
could potentially be significantly worse than indicated at later times.
5.2.3 Practical Considerations
Prior to drawing conclusions and making recommendations about the allowed
carriage of baggage, it is important to determine what possibilities for policy change
are realistic and which are unrealistic or impossible. The simplest method for mea-
suring and meeting the current Coast Guard policy requirements would be to weigh
each bag brought aboard the vessel and require that all baggage be less than the
77
maximum weight per person calculated using the seat density of each particular ves-
sel. This would place an unreasonable restriction on baggage, as very few bags are
under the 1.4 kg (3.1 lb) weight restriction necessary for high seat density vessels.
This measure would also be impossible for companies to enact as the time required
to weigh each piece of baggage would prevent ferry vessels from running on a regular
or predictable schedule.
If the weight of the baggage cannot be effectively managed, the number of
bags brought aboard is another option. Limiting the number of bags carried by each
person, or perhaps limiting the relative size of the baggage, would allow a reasonable
and possible limitation on the transient fire load place in the compartment. This
measure could only be effective and realistic if the ferry vessel was able to provide
additional compartments specifically for the storage of baggage that is not allowed
into the passenger compartment. As such, this measure would create additional
burdens on the company, but is potentially realistic from an operations standpoint.
One final possibility is to alter the storage of the baggage within the com-
partment itself. Eliminating stowage of strollers or bulky items in compartment
corners, and preventing storage of baggage on top of seats are two possible options.
For the present study baggage was considered to be placed on top of seat cushions,
the likeliest location based on vessel surveys. If however, the baggage were placed
underneath seats, or in special storage compartments between aisles, the potential
for fire size may be decreased. Baggage stored underneath the seat would limit the
height of the fire plume, and thus result in lower ceiling temperatures. In addition
the decreased height of the fire could have a positive impact in limiting the spread
78
of the fire across knee gaps and aisles. However; this option would require major
modifications to passenger compartments to ensure adequate stowage options, as
well as additional training for crews and lengthy re-education of passengers who are
accustomed to a high degree of behavioral freedom on ferry vessels as compared to
an airplane.
79
Chapter 6: Conclusions and Recommendations
6.1 Conclusions
It is clear that the current level of carriage aboard aluminum passenger ferries
with type 5A spaces presents a significant danger in terms of increasing the fire load
within the space. It is relatively simple to calculate and control the fixed fire load
within a space, but controlling the baggage and baggage content brought aboard by
passengers is neither simple nor straightforward. Although some companies have
strict baggage policies in place, these policies are not fully able to account for the
weight of every piece of baggage brought aboard, and as a result Coast Guard
policy requirements are not strictly met. Survey data indicates that the current
average baggage weight of 3.7 kg exceeds that allowed by Coast Guard policy for
93% of vessels, with the remaining 7% falling within the policy requirements due to
unusually low seat density in the main passenger compartment.
The weight of baggage brought aboard ferry vessels has the potential to raise
the temperature of the aluminum ceiling above 200?C in less than 12 minutes when
a fire occurs with baggage left on top of seats. The aluminum ceiling reaches a
temperature of 253?C, corresponding to a loss of more than 50% of the structural
integrity of the aluminum and creating the potential for a lethal structural failure.
80
This structural failure would endanger the lives of passengers that may be gathered
on the deck above in preparation for abandoning ship. In addition there are at
least two further scenarios, a stroller parked in the corner of the compartment and a
carry-on bag in the middle of an escape aisle, that also create dangerous situations
and exceed current temperature limitations. In the case of the stroller on fire in the
corner of the compartment the danger is exceedingly great as the melting tempera-
ture of aluminum is exceeded for a period of two and a half minutes, which would
assuredly result in significant damage to the vessel?s superstructure.
6.2 Recommendations
A simple solution to the problem would be to limit the weight of baggage
each person is allowed to bring aboard to 1.4 kg (3.1 lbs) per person, the value
that corresponds to the weight limit per area of 2.5 kg/m2 for the vessel with the
greatest seat density. However, this would unnecessarily restrict vessels with low
seat densities and would place an unreasonable restriction on baggage weight by
eliminating the carriage of 96% of bags currently brought aboard vessels. In addition
the effort and time of weighing each individual bag as it comes on board would
severely impact the current business procedures of ferry companies who depend on
a swift unloading and reloading in order to maintain tight schedules. It would also
be unrealistic to expect to eliminate transient fire load altogether as passengers
have a reasonable expectation that they will be able to carry at least some personal
belongings with them, particularly those belongings that may be necessary for use
81
during transits.
A more practical solution may be to limit baggage to one small size bag per
person. Persons that attempt to bring aboard multiple bags or larger items such as
suitcases (even carry-on size suitcases) could be limited to one small bag with which
to carry valuables and important belongings, while having remaining baggage stored
in a sprinkler protected luggage compartment. For items in contention, a weight
limit of either 4.5 kg (10 lbs) or 6.8 kg (15 lbs) could instituted. The decrease in
the weight of the average bag associated with imposing these weight limits would
mean that 28.8% of vessels would be in full compliance with Coast Guard policy
requirements with the limit set at 6.8 kg (15 lbs) and 62.7% of vessels would be
in full compliance with Coast Guard policy with the limit set at 4.5 kg (10 lbs).
In addition it would be necessary to make provisions for large bulky items such
as strollers so that these items would not be placed in corners, along walls, and
in escape aisles where they could inhibit evacuation and result in dangerous fire
scenarios. Finally, storage of all items could be strictly eliminated in problem areas
such as corners, against walls, and in escape aisles.
82
Appendices
83
Appendix A: Vessel Survey Data
Additional data from the vessel surveys is provided here; including histograms
for individual bag types, weight data from measured bags, and carriage distribution
graphics.
84
Figure A.1: Baggage carriage distribution for ferry vessels.
Figure A.2: Baggage carriage distribution including only small bags car-
ried into the passenger compartment.
85
Figure A.3: Baggage weight histogram for all baggage combined.
Figure A.4: Baggage weight histogram for purse.
86
Figure A.5: Baggage weight histogram for backpack.
Figure A.6: Baggage weight histogram for small duffel bag.
87
Figure A.7: Baggage weight histogram for large duffel bag.
Figure A.8: Baggage weight histogram for carry-on size suitcase.
88
Figure A.9: Baggage weight histogram for briefcase.
Figure A.10: Baggage weight histogram for tote bag.
89
Figure A.11: Average weights of each specified baggage type.
Figure A.12: Comparison between average weights from Swedish Study
[3] and vessel survey data.
90
Figure A.13: Measured weight data from all baggage weighed during
vessel surveys.
91
Appendix B: Baggage Heat-Release Rates
Linear heat-release rate curves were used to create an average heat-release rate
curve for a piece of passenger baggage in FDS simulations. Figures B.1 - B.6 show
the curve for a stroller, carry-on, briefcase, purse, small duffel, and backpack.
92
Figure B.1: Liner heat-release rate curve use for a stroller based on data
from Kumm [3].
93
Figure B.2: Liner heat-release rate curve use for a carry-on bag based
on data from Kumm [3].
94
Figure B.3: Liner heat-release rate curve use for a briefcase based on
data from Kumm [3].
95
Figure B.4: Liner heat-release rate curve use for a purse based on data
from Kumm [3].
96
Figure B.5: Liner heat-release rate curve use for a small duffel bag based
on data from Kumm [3].
97
Figure B.6: Liner heat-release rate curve use for a backpack based on
data from Kumm [3].
98
Appendix C: FDS Simulation Results
This appendix provides additional images and figures depicting the results
from FDS simulations for the three tested scenarios.
99
Figure C.1: Single point ceiling temperature above the burning stoller.
100
Figure C.2: Ceiling temperature above burning stroller in the corner of
the compartment.
101
Figure C.3: Maximum single point ceiling temperature above a single
bank of burning seats.
102
Figure C.4: Maximum single point ceiling temperature above two burn-
ing seat banks.
103
Figure C.5: Average ceiling temperature over one square meter above
two burning seat banks.
104
Bibliography
[1] AZoNetwork UK Ltd. Aluminium alloys - aluminium 6082 properties, fabrica-
tion and applications, supplier data by aalco. 7 2013.
[2] Sujeebun. S Manohar Wong Ky. B Sblin. B, Jahazeeah. Y. Material science -
steel. 2013.
[3] Mia Kumm. Carried fire load in mass transport systems. Studies in sustainable
technology, Ma?lardalen University, Va?ster?as, Sweden, 2010.
[4] 16 dead in fatal fire on indonesian ferry. CBC News, February 2007.
[5] Chester M. Sprague P.E. and Brian L. Dolph. Low fire load compartment
flashover testing. Technical Report CG-D-05-98, U.S. Coast Guard Research
and Development Center, Groton, CT, December 1997.
[6] Steamshipauthority.com vessels, December 2013.
[7] John Duclos. Private Communication, September 2013. Gladding-Hearn Ship-
building, The Duclos Corporation.
[8] Noel Shriner. Fire growth evaluation for regulations of fire load for type 5a
spaces on seafaring vessels. Master?s thesis, University of Maryland, Colleg
Park, College Park, MD, 2012.
[9] Pierre Palmberg and Henrik Georgsson. Scandanavian star fire investigation.
Technical Report 4000455000, SWECO Fire Protection Engineering and Risk
Management, Malmo?, Sweeden, 2009.
[10] Reynaldo Garibaldi and Napoleon Smith. Preliminary investigation report on
the sinking of m/v al salam boccaccio 98. Technical report, Panama Maritime
Authority, Panama, August 2006.
[11] Sector Southeastern New England. Investigation activity report: Lady martha.
Technical Report 3374272, U.S. Coast Guard, East Providence, RI, September
2008.
105
[12] Hyannis firefighters quickly extinguish fire aboard the lady martha. Cape Cod
Today, September 2008.
[13] Marine Technical Chief and Hazardous Materials Division. Policy File Mem-
orandum on Structural Insulation Requirements for Low Fire Load Spaces on
Certain Vessels. U.S. Coast Guard, Washington, D.C., 1994.
[14] Navigation and Vessel Inspection Circular 9-97. Guide to Structural Fire Pro-
tection. U.S. Coast Guard, Washington, D.C., 2010.
[15] Simo Hostikka Kevin McGrattan, Randall McDermott and Jason Floyd. Fire
Dynamics Simulator (Version 5) Users?s Guide. National Institute of Standards
and Technology, Washington, D.C., 2010.
[16] Chester M. Sprague P.E. and Brian L. Dolph. Theoretical basis of the ship
fire safety engineering methodology. Technical Report CG-D-30-96, U.S. Coast
Guard Research and Development Center, Groton, CT, September 1996.
[17] Capt. Greg Gifford. Private Communication, July 2013. Steamship Authority.
[18] Christian Myers. Private Communication, August 2013. Interstate Navigation.
[19] Marine Safety Center Technical Note No. 01-13. Application of Low Fire Load
(Type 5A) Space Policy. U.S. Coast Guard, Washington, D.C., 2013.
[20] 14 CFR 121. Flight Operations, Operating Requirements: Domestic, Flag, and
Supplemental Operations, 2014.
[21] Federal Aviation Administration. Carry-On Baggage Tips, Preparing to Fly,
2013.
[22] G De Matteis-P Haller A Santiago L Simoes da Silva P Vial Real F Wald,
V Bosiljkov. Structural integrity of buildings under exceptional fire. COST
C-12 Seminar, 2002.
[23] National Transportation Safety Board. Emergency evacuation of commercial
airplanes. safety study ntsb/ss-00/01. Technical report, NTSB, Washington,
DC, 2000.
[24] Jin-Hyoung Park-Beom-Jin Park Dongkon Lee, Hongtae Kim. The current
status and future issues in human evacuation from ships. Safety Science, 41,
2003.
[25] S Deere-L Filippidis R Brown, E R Galea. Passenger response time data-sets
for large passenger ferries and cruise ships derived from sea trials. International
Journal of Maritime Engineering, 2013.
[26] Robert Brown Lazaros Filippidis Edwin Galea, Steven Deere. An experimental
validation of an evacuation model using data sets generated from two large
passenger ships. Journal of Ship Research, 57:1?16, 2013.
106