446 J. AIRCRAFT VOL. 6, NO.

Analyses of Passenger and Baggage Flows in Airport

Terminal Buildings
ROBERT HORONJEFF*
Institute of Transportation and Traffic Engineering, University of California, Berkeley, Calif.

Passenger flows into departure lounges can be expressed by an S-shaped quartic curve.
The flow of passengers from the departure lounge into an airplane and from the airplane
into the lounge is nearly linear. Scheduling of arrivals of large aircraft at about the same
time will result in very high passenger flow rates in pier finger corridors during very short
intervals of time. Deterministic theory can be applied to passenger and baggage arrival
characteristics which in turn can be used to estimate space requirements for passengers in
baggage claim area.
Downloaded by UNIVERSITY OF MICHIGAN on March 1, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.59431

Introduction models were developed from observations at San Francisco

International Airport; therefore, no claim is made for their
T HE air transport system is essentially made up of three
major subsystems: 1) access to the airport, 2) processing
at airports, and 3) flight. This paper deals with processing
validity elsewhere. They merely indicate a type of approach
that can be made for making design decisions concerning the
sizing of departure lounges.
at the airport, with emphasis on passengers and baggage. The common procedure followed by a number of airlines is
In terms of research, the effort devoted to the flight and to commence checking passengers into the lounge approxi-
access subsystems has been much greater than that devoted mately 60 min prior to scheduled departure time and to open
to processing at the airport. The reasons for this are under- the door to the jetway 10-20 min prior to departure time.
standable. All of the activities related to flight are under This procedure was used at San Francisco International Air-
the jurisdiction of or are of direct interest to the Federal port during the time this study was made; therefore, the
Government; hence, there has been substantial Federal sup- study results are only applicable if similar procedures will be
port in this area. Likewise, a good share of the access to used in the future.
airports has been by automobile, and the entire street and The flights sampled were all long-range—to New York,
highway program has received substantial support for re- Honolulu, Washington, St. Louis, and Philadelphia—so the
search from the Federal and State Governments. But be- characteristics of commuter flights are not included. The
tween those two areas lies the relatively unexplored area of flow of passengers to the departure lounge is time-dependent,
passenger and baggage flows through the terminal building. and also dependent upon the activities of the passengers
The prime responsibility for the design of the terminal build- elsewhere in the terminal. The total number which arrive
ing rests with the airport owner, who does not have the re- at the lounge at any given time before departure can be
sources to invest in research. In recent years, the airlines called the cumulative flow. This is illustrated in Fig. 1.
have undertaken to develop passenger processing techniques The cumulative flow is F(t), where t is minutes before de-
to deal with growing volumes of traffic, but the body of parture. The total number of passengers waiting in the
knowledge available to date is far less than for the other lounge at any time t is Q(t). The time that the entrance
major subsystems. doors into the aircraft are opened is designated as fa and the
The introduction in the 1970's of large jet aircraft capable flow into the aircraft is G(t). Before fa, the number of
of carrying several hundreds of passengers will create new passengers waiting in the lounge Q(t) is equal to the cumula-
space requirements in airport terminal buildings. A better tive flow to the lounge. Between fa and £2, Q(t) is equal to
understanding of the flow processes in these buildings will the difference between F(t) and G(t). At time t2, the queue
ultimately be useful in making decisions concerning their dissipates and the flow rate into the aircraft thereafter is
design. With this in mind, we have been applying some equal to [dF(t)]/(dt). If the capacity flow rate into the air-
research effort in this direction. craft is always equal to or greater than the flow rate into

Passenger Flow to Departure Lounges

Space for departure lounges will be a problem with the
introduction of large jets. It was decided to study the char-
acteristics of passenger flows to departure lounges and from
the departure lounge into the aircraft. From this analysis
some clues to the interrelationship between the size of the
lounge, the number of aircraft doors available for loading,
and the time allotted to loading the aircraft were developed.
Two analytical models were developed; one model describes
the flow process of passengers into the departure lounge and
the other the flow from the lounge into the aircraft. The

Presented as Paper 68-111 at the AIAA 5th Annual Meeting

and Technical Display, Philadelphia, Pa., October 21-24, 1968; Time (minutes before departure)
submitted October 21, 1968; revision received February 7, 1969.
* Professor of Transportation Engineering. Member AIAA. Fig. 1 Typical boarding sequence.
 SEPT.-OCT. 1969 PASSENGER AND BAGGAGE FLOWS 447

the lounge, the maximum will occur at some time after U, i.e., © b = 14 Passengers per minute (one door /
® b=28 Passengers per minute (two doors)
nearer to departure time. Models were developed for F(t) © b=42 Passengers per minute (three doors)
and G(t). ® b =56 Passengers per minute (four doors)
The model for F(t) was determined by observing the flow
to the lounge. Passengers were counted as they arrived
and their arrival time was recorded. This procedure was
continued until departure time. Prior to developing the
model, the number of passengers on each flight was normal-
ized in terms of percent. This step served as the basis for
using the model to predict the number of passengers arriving
at any time t, if the total number of passengers to be boarding
a given flight was known.
60 50 40 30 20 IO 5 O
The data obtained were used in a least-squares regression Minutes Before Departure
analysis. It was found that the best fit, i.e., the curve having
a minimum value of the sum of the squares of the deviations Fig. 3 Departure lounge sizes for 300 passenger airplanes.
of the theoretical points from the observed points, was a
quartic curve of the form An average capacity flow rate was calculated from the ob-
2
^(0 = GO + ait + a2t + a^ + a±t* served flights, and the value of b turned out to be 14. The
final model for passenger flow into an aircraft was, therefore,
where F(t) = percent of passengers boarding a flight, t =
Downloaded by UNIVERSITY OF MICHIGAN on March 1, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.59431

minutes before departure, ao = —1.78, a\ — 0.72, 0,2 = —0.02, Q(t) = U(t - tb)
a3 = 0.0025, and a4 = -0.00003. Ninety-nine percent confidence limits for the expected value
The standard error of estimate of F(t) on t was found to be and variance of the capacity flow rate were 12 < £ < 14 and
<7 = 1.6. A chi-square test for goodness of fit of the curve 1.3 < o-2 < 17.8, respectively.
to the data revealed that the fit was good with a probability The models can be used to investigate the relationship
of 0.55. Detailed calculations are contained in a study by between the number of aircraft doors, the time before de-
Paullin.1 (60-0 is used as abscissa in Figs. 2, 3, and 4. parture that these doors are opened, and the size of the de-
The model for G(t), the flow into the airplane, was developed parture lounge. Tradeoffs in time and space can be analyzed
by observing passenger flow through the jetway and aircraft and decisions formulated regarding future loading procedures.
door. A regular pattern was evident, in that all or most of Applications of the models are shown in Figs. 2, 3, and 4.
the waiting passengers began queueing in the jetway as soon From these illustrations, it is evident that the size of a de-
as the boarding announcement was made. The queue lasted parture lounge depends on 1) the size of the aircraft, 2) the
until waiting passengers were boarded, at which time the number of available entry doors into the aircraft, 3) the
flow into the aircraft reduced to equal the flow of passengers arrival pattern of passengers to the lounge, and 4) the time
arriving at the lounge or G(t) — F(t). allowed for boarding passengers. Increasing the number of
During the period of queueing in the jetway, flow rate into entry doors into the aircraft permits a reduction in boarding
the aircraft was maximum. The period of queueing varied time; but if this is the goal, it is achieved at the expense of
from 3 to 5 min, during which the flow rate was nearly con- increasing the size of the departure lounge.
stant. The model for the flow into the aircraft was, there-
fore, selected as
Additional Observations on Inflow and Outflow
G(t) = b(t - tb<t<t2 of Passengers
where G(t) = number of passengers having boarded at time Since the completion of the study by Paullin,1 additional
t,b = capacity flow rate, passengers/min, t = minutes before observations were made on rates of flow of not only enplaning
departure, tb = initial boarding time, and fa = time at which but also of deplaning passengers.2 Passenger flow character-
queue dissipates (see Fig. 1). istics and estimated flow rates for loading passengers were
It should be noted that G(t) is absolute, in terms of number determined for two passenger handling strategies. The first
of passengers and not in percent of passengers as is F(t). The strategy was to load passengers without seat assignment and
term G(t) is a function of the size and number of aircraft doors the second was to load passengers with seat assignment. In
available for loading and the interior configuration of the addition, unloading rates were also observed. All passengers
cabin. were loaded and unloaded through a jetway. A queue of
passengers existed for all of the observed flights. The

IO, 20,30 minutes before departure

60 50 40 30 20 IO
60 5O 4O 3O 2O IO O Minutes Before Departure
Minutes Before Departure
Fig. 4 Departure lounge sizes for B-747, 500 passenger
Fig. 2 Passenger arrivals at departure lounge. capacity.
 448 R. HORONJEFF J. AIRCRAFT

Passenger f/o

Baggageunloadarea
bags
Q passengers

Fig. 8 Baggage claim system.

L o ading Time -
pass, etc.), 4) aisle length and width, 5) seat assignment vs
Fig. 5 Cumulative number of enplaned passengers no seat assignment, 6) number of seats/row served by an aisle,
(passengers loading with seat assignment-—Boeing 727). 7) seat spacing (longitudinal), and 8) composition of passen-
gers. Only item 5 above was studied in detail by Kaneko.2
flights sampled were primarily short-haul domestic flights But even from this very preliminary study, it was clearly
to and from San Francisco International Airport. evident that the constraint in loading was not the size of the
Typical results for enplaning passengers are shown in aircraft door but the amount of congestion in the aisle.
Figs. 5 and 6. It will be noted that 1) enplaning rates do not
remain constant throughout the loading time but that an Passenger and Baggage Flow at Arrival
Downloaded by UNIVERSITY OF MICHIGAN on March 1, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.59431

initial surge occurs and 2) the rate is higher for flights with Baggage Claim Area
no seat assignment. With seat assignment, the average
loading rate at 4 min was about 16 passengers/min, which is Space requirements for baggage claim are an important
not much different than the 14 passengers/min reported by input in the design of airport terminal buildings. Space
Paullin.1 With no seat assignment, however, the average needs can be greatly influenced by the interrelationship of
rate was about 20 passengers/min. passenger and baggage arrival patterns at the baggage claim
Of equal importance is the unloading rate, since the number area. A deterministic queueing model was developed to re-
of doors required for large aircraft could well be governed by late the number of bags on a carousel to the arrival distribu-
the time desired to empty rather than load an aircraft. Fig- tion of passengers and baggage. The model was based on
experimental data taken at San Francisco International Air-
port.3 From these data, the following were developed:
1) cumulative passenger arrivals at the baggage claim at
Y2 = 11+17tf.
time t = Ap(t), 2) cumulative bag arrivals at the carousel
at time t = Ab(i), and 3) bags remaining unclaimed at select
times after the start of baggage arrival. Time was measured
- A UAL5II- from the instant an arriving flight began to disembark
Y, =24t,
° UAL5I9
• UAL 521
passengers.
o UAL 523
The baggage claim system at San Francisco International
Airport is schematically illustrated in Fig. 8. When the air-
craft reaches its position on the ramp, personnel move a
Loading Time-minutes train of baggage trailers into position at the aircraft. For
Fig. 6 Cumulative number of enplaned passengers
containerized aircraft, containers are lowered onto gondola
(passengers loading without seat assignment—Boeing 727). trailers, which are then towed to the baggage unload area for
unloading. For pit loaded aircraft, bags are unloaded, with
the help of mechanized loading equipment, to flatbed trailers.
ure 7 illustrates the typical results from several flights and
Airport policy limits the number of trailers in tow to six, and
indicates that the rate is fairly constant at about 36 passen- the airlines frequently tow fewer than this number. Thus,
gers/min.
for large baggage loads, several trips may be required be-
Paullin1 and Kaneko2 both point out that the rates of flow
tween the aircraft and baggage unload area.
are for the present size of doors. However, the doors on the At the baggage unload area, ramp crews manually remove
747 and other aircraft will be larger, so that flow rates will bags from the trailers and place them on a conveyor belt
probably be greater. At no time during the observations
leading to a carousel. Maximum flow rate for the belt,
were the jetways a constraint, either for loading or unloading.
basedjon the capacity of the carousel to accept bags without
There are many factors that can affect passenger flow clogging, is approximately 40 bags/mm. The unloading
rates into an aircraft. Unfortunately, there was no oppor-
rate for one man varies from 10-15 bags/mm. Thus, the
tunity to study all of them, but they are listed below in the
delivery rate can be very dependent on the size of the crew
event continuing studies are made. These factors are as offloading baggage.
follows: 1) width of jetway, 2) aircraft entry door width,
3), type of service rendered at aircraft door (collect gate

<f 6 8 / 0 / 2 / 4 16
Time in Minutes After Aircraft Block Time
Unloading Time - minutes
Fig. 9 Application of analysis to data of flight AA 225,
Fig. 7 Cumulative number of deplaned passengers. July 28, 1967.
 SEPT.-OCT. 1969 PASSENGER AND BAGGAGE FLOWS 449

6 8 IO 12 14 16 18 4 6 8 IO 12 14 16 18 20
Time in Minutes After Aircraft Block Time Time in Minutes After Aircraft Block Time

Fig. 10 Application of analysis to data of flight TWA 41, Fig. 12 Analysis of baggage requirements for a flight with
July 20, 1967. 200 passengers and 400 bags at 40 bags/min delivery rate—
delayed delivery.
Flights known to have large passenger and baggage load-
ings were selected for analysis. The structure of the analysis To compute Wi, the following logic was used. There is
is shown in Fig. 9. Knowing Ap(t), the probability dis- valid reason to assume that the arrival time of a bag Tb and
tribution of a passenger arriving at the baggage claim area the arrival time of the passenger who seeks his bag TP are
Downloaded by UNIVERSITY OF MICHIGAN on March 1, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.59431

Fp(t) = Ap(t)/Np < 1, where Np is the total number of independent random variables. The time at which both
arriving passengers. Knowing Ab(t), the probability dis- passenger and his bag have arrived at the carousel is Tp,b =
tribution of a bag arriving at the carousel is Fb(t) = Ab(t)/Nb max (Tp,Tb), and the probability distribution function of
< I , where Nb = total number of bags. Since the bags Tp<b is then Fp,b(t) .= Fp(t)Fb(t) which is the probability that
remaining on the carousel were counted for select times, the both the passenger and his bag arrive at the carousel at time
expression for the cumulative number of bags removed from t. The average waiting time Wi is the area between the two
the carousel at time t, Db(f), equals Ab(t) minus bags on carou- curves Fb(t) and Fp(i)Fb(t), which is
sel at time t. Then the probability distribution of a bag re-
moved from the carousel is Fd(t) = Db(t)/Nb < 1. At any \l* [Fb(t) - FP(t)Fb(t)]dt
time t, the total number of bags remaining on the carousel
Qb(t) is Nb[Fb(t) - Fd(t)], where Nb = total number of bags
on a flight. This total should not be much larger than the
capacity of the carousel, which at San Francisco was about
80 bags (one-row deep—25.5 ft in diameter—circumference
80ft).
The average waiting time W for a bag to remain on a carou-
sel is the area between the two curves Fb(t) and F(i(i), which
is equal to

','' Fb(t)Fd(t)dt
4 6 8 10 12 14 16 18 2O
where ti = arrival time of the first bag and ti is the last bag
Time in Minutes After Aircraft Block Time
to be claimed. W is the shaded area shown in Fig. 9 multi-
plied by the total number of bags Nb. This average waiting Fig. 13 Analysis of baggage requirements for a flight
time is the sum of two time periods W\ and Wz- The first with 200 passengers and 400 bags at 30 bags/min de-
period W\ is the average waiting time when the bag arrives livery rate.
at the carousel first and must wait for the arrival of the
passengers and Wz is the average waiting time for the passen- The average waiting time Wz was determined experimen-
ger to remove the bag from the carousel, assuming that both tally. A reasonable estimate of W^ would be one-half the
passenger and bag are at the carousel. Clearly, W\ would time for the carousel to make one revolution, this being
be equal to zero if all passengers arrived before the first bag J min at San Francisco. This estimate checked closely with
arrived. If the passenger is at a certain position along the experimental data for flights with few passengers and bags,
circumference of the carousel, waiting for his bag and the but for heavily loaded flights the value of Wi was closer
bag is at the mouth of the carousel, Wz would be the average to 1 min.
time taken for the bag to travel from the mouth of the carousel In summary, the mechanics of the analysis are as follows:
to the location of the passenger and W\ is equal to zero. determine Fp(t) and Fb(t) experimentally. Compute FP(t)-
Fb(t) and then displace this line by an average value of
Wi(%-l min) and obtain Fd(t).
To verify the accuracy of the analysis, compare the com-
puted Fd(t) with the observed Fd(t) as shown in Figs. 9 and
10. Figure 9 illustrates the application to a specific flight,
AA225, and Fig. 10 the application to TWA Flight 41.
Figures 11, 12, and 13 show some applications of the anal-
ysis in comparing strategies for baggage delivery. In each

Aircraft
6 8 IO 12 14 16 18 2O Terminal Pier finger corridor — Gates
Building
Time in Minutes After Aircraft Block Time

Fig. 11 Analysis of baggage requirements for a flight Entrance of corridor to satellite

with 200 passengers and 400 bags at 40 bags/min delivery
rate. Fig. 14 Satellite arrangement.
 450 R. HORONJEFF J. AIRCRAFT

FROM AIRCRAFT Table 1 Forecast of aircraft schedules

• Arriving
^H Passengers No. of No. of
entrance passengers
/^\ Aircraft and exit for each
FROM ——————————————
TERMINAL ^ Departing Passenger Arriving Passengers FLOW ^ type doors flight Arrival Departure
T0
FLOW ^ Arrival Grueters Departure Visitors
X / ___________ TERMINAL DC-8 1 100 1645 1800

V^ ^H 00po/
• Passe
DC-8
B-727
B-727
1
1
1
100
76
76
1705
1730(0)
1745
&
1830
1800
1900
B-727 1 76 1745 1830
TO AIRCRAFT
DC-8 1 100 1755 1855
DC-8 1 100 1755 1855
Fig. 15 Passenger flow model. B-727 1 76 1800 l820(T)a
DC-8 1 100 1800 1900
case, Fp(t) is based on a United Air Lines DC-8-61 with 200 DC-8 1 100 1805 1855
passengers, and Wi is equal to 1 min. B-727 1 76 1810 1900
DC-8 1 100 1815 1930
The comparison of Figs. 14 and 15 shows the effect of the B-747 4 312 1815 1900
start of delivery time on bag storage requirements. In each B-727 1 76 1830 1915
case, delivery rate is 40 bags/min, the maximum rate for B-727 1 76 1840 1935
existing carousels. Note that for a strategy of early bag
Downloaded by UNIVERSITY OF MICHIGAN on March 1, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.59431

B-727 1 76 1845 2000

delivery, storage equipment Qb(t) is 148 bags compared to 85 DC-8 1 100 1855 1920(T)a
bags for a strategy of delivery after passengers start arriving. B-727 1 76 1900 2030
Note also that there is little savings in total time for early T = termination flight.
bag delivery. b
0 = origination flight.
The strategy for this flight (Fig. 12) was to provide all
bags in the shortest possible time. Arrival baggage rates terminal building by a corridor. In this study, no aircraft
at times exceeded 90 bags per minute. The number of bags is assumed to be parked along the side of the corridor. The
waiting to be claimed often exceeded 150 which is what one model develops the flow of people at the entrance of the
would predict from the analysis. corridor to the satellite. The characteristics of flow within
Comparison of Figs. 11 and 13 provides information on the the corridor are the subject of another study.
effect of reducing baggage delivery rate. Figure 13 illus- The types of people using the corridors are 1) departing
trates a delivery rate of 30 bags/min compared to 40 bags/ passengers, 2) visitors accompanying departing passengers,
min in Fig. 11. The starting time of delivery is the same for 3) arriving passengers, 4) greeters accompanying arriving
each case. The lower flow rate does reduce storage require- passengers, 5) sightseers, and 6) employees. Sightseers and
ments slightly; however, the tradeoff is an increase in total employees were excluded in the initial model application be-
carousel occupancy time. cause of lack of data on their number; however, the model
permits their inclusion whenever the data are available.
Passenger Flow in Pier Fingers of For the purpose of modeling the passenger flow process, the
Airport Terminals configuration shown in Fig. 15 was assumed. The process
is divided into two separate and distinct operations, arrivals
The movement of people in corridors of airport terminals and departures.
is a vital element in their design. Particular questions that In developing any model, simplifying assumptions must
need to be answered are 1) what are the combined passenger be made at the start. After the initial effort, the model must
and visitor flows in corridors of pier fingers that will be gen- be tested in a real life situation. At this point in time, one
erated in the future and 2) what should the width of the can decide to what extent the initial model should be modi-
corridors be to accommodate these flows? fied. The assumptions that have been made for the initial
The objectives of the study described by Smith4 were 1) model are summarized as follows: 1) the aircraft arrival and
to develop a simulation model which could identify the im-
portant parameters influencing the flow rates in a corridor
generated by arriving and departing aircraft clustered
around a satellite and 2) hopefully, to predict within reason
what these total flows would be. The term "satellite" refers
to a circular area in which the departure lounges are housed,
as shown in Fig. 14. The satellite is connected to the main

16:4017:00 :20 :40 18:00 :20 .40 13:00 :20 *0 21:00 10 20 X 40 50 60 TO dO 9O /OO IrO J2O
Time of Day

Fig. 16 Scheduled aircraft gate occupancy. Fig. 17 Flow of people in pier finger corridor.
 SEPT.-OCT. 1969 PASSENGER AND BAGGAGE FLOWS 451

departure times from the gates are the same as opening the The model has been made sufficiently flexible to accept
doors to arrivals and closing doors for departures, 2) the early or late aircraft arrivals and late departures (from
flows of people in opposite directions do not interact signifi- scheduled). In this way, one is able to examine the conse-
Downloaded by UNIVERSITY OF MICHIGAN on March 1, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.59431

cantly, i.e., the corridor and the satellite area are wide enough quences of not being on schedule on 1) passenger flows and
so that free flow can occur in each direction, and 3) the time 2) delay to passengers due to lack of available gate positions.
to walk from any jetway or departure lounge to the entrance It was assumed that arrivals could be as early as 10 min and
of the corridor is the same for all gates. as late as 20 min, and that between these two extremes the
In developing the simulation model, the generation of flow probability of such an occurrence was uniformly distributed
in the pier ringer was assumed as follows. (equal probability). Minimum service times were obtained
for several types of aircraft, and these were applied to the
Generation of Departing Passengers and Departure late arrivals whenever the lateness was sufficient to delay
the scheduled departure time.
Visitors (DEPGEN) To illustrate the application of the model, a hypothetical
1) Departing passengers are assumed to arrive at the case was investigated. Given the following schedules (Fig.
departure lounge in accordance with the investigation made 16 and Table 1), aircraft load factors, number of jetways,
by R. Paullin described earlier in this paper1 and by Paullin and number of visitors, and assuming that the aircraft are all
and Horonjeff. 5 on time, what would the people flow be at the entrance of the
2) Departure visitors are assumed to accompany departing
corridor to the satellite?
It was assumed that for this particular application there
passengers.
were -f- visitors for each passenger. The results of the simula-
3) Departure visitors are assumed to leave the departure
tion are shown on Fig. 17.
area in accordance with the S-shaped cumulative curve de-
veloped by R. Paullin1 but compressed in time to a total of
References
10 min. It was also assumed that the first visitor leaves the 1
satellite 5 min before the departure of the aircraft. Paullin, R. L., "Passenger Flow at Departure Lounges/'
Graduate Report, Institute of Transportation and Traffic Engi-
Generation of arriving passengers and arrival neering, Univ. of California, Berkeley, July 1966.
2
greeters (ARGEN) Kaneko, E. T., "Passenger Enplaning and Deplaning Char-
acteristics/' Graduate Report, Institute of Transportation and
1) Arriving passengers are assumed to exit from the air- Traffic Engineering, Univ. of California, Berkeley, Aug. 1967.
3
port at a constant rate for each jetway (aircraft flow) based Barbo, W. A., "The Use of Queueing Models in Design of
on a study by Kaneko and others2 and leave the satellite Baggage Claim Areas at Airports/' Graduate Report, Institute
at this rate. of Transportation and Traffic Engineering, Univ. of California,
2) Arrival greeters are assumed to accompany arriving Berkeley, Sept. 1967.
4
passengers when leaving the satellite. Smith, E. E., "Simulation of Passenger Flows in Pier Fin-
gers," Graduate Report, Institute of Transportation and Traffic
3) Arrival greeters are assumed to arrive at the satellite Engineering, Univ. of California, Berkeley, Aug. 1968.
with a distribution similar to the departure passengers but 5
Paullin, R. L. and Horonjeff, R., "Sizing of Departure
compressed in time to a maximum of 15 min. Lounges in Airport Terminal Buildings/' Proceedings ASCE-
It was also assumed that the last greeter arrived at^the AOCI Specialty Conference, Houston, Texas, April 10-14,
same time as the aircraft. 1967.