Professional Documents
Culture Documents
The F-110 frigates are intended to be new and complex communication between the different operating positions.
weapon-equipped vessels integrating advanced systems that This is regarded as critical for the success of the operations
allow the Spanish Navy to cover operational and strategic in which a naval vessel is involved.[1]
missions for the defense and security of the country and its
allies. However, the main novelty of the F-110 series will Combat Information Center (CIC)
consist of an integrated mast, where the communication sys- The primary mission of the CIC is to provide organized col-
tems, the radar, the electronic warfare and the optronics will lection, processing, display, competent evaluation, and rapid
be located together so as to avoid rotating elements.[1] dissemination of pertinent tactical information and intel-
The new F-110 frigates must also be designed to be ligence to command and control stations.[2] The CIC is re-
versatile and balanced ships in all areas of naval warfare; i.e, sponsible for keeping the Commander permanently advised
they must be able to operate in both high and low-intensity of the current tactical situation and recommending the most
scenarios. Thus, their design should respond to the following appropriate action to be taken. The CIC is also known as the
objectives: Operations Room, and it is present in all kind of warships.
■■ Capability to perform in both open-ocean and littoral coast The functions of the CIC can be summarized as follows:
operations ■■ Gathering information from sensors in the proper vessel
■■ High survivability (reduced probability of being detected panels or video screens (Figure 3).
and notable resistance against possible damages) ■■ Evaluating the available information and proposing the
■■ Balanced combat system in all war segments: AAW, ASUW, optimal operational decision.
ASW, electronic warfare, asymmetric defense ■■ Communicating information/commands to the differ-
■■ Modularity and flexibility (multimission bay) ent control stations and other fleet ships.The operations
The F-110 project is currently in the conceptual design within the CIC are grouped according to their combat
and definition phase, where the final layout of the CIC roles and functions. Every system specialist has their own
equipment and consoles has to be defined. One simple console to operate and control the assigned sensor or
solution could be adapting the CIC arrangement of the F-100 weapon. The current design trends in the provision of the
frigates to the new ship characteristics. However, after the CIC equipment can be simplified into three broad types:
experience gained in more than 15 years of operational use ■■ Command panel in the center: CENTER
and taking into account the design trends observed in other ■■ Command panel on one side: SIDE
navies and the evolution of the conflict scenarios towards ■■ NASA type command panel
validate it, a real evacuation experiment was carried out. Lastly, the obtained results were analyzed and global
Data generated by computer simulation were an adequate conclusions were drawn.
approximation to the experimental framework of the
problem. Multiple criteria decision-making
■■ Step 5. Evacuation time responses were obtained with Decision-making can be defined as the choice, on some basis
Pathfinder simulation software according to NATO Naval or criteria, of one alternative from a set of them. A decision
Ship Code, showing that factors level variation did not may need to be taken based on multiple criteria rather than
produce a mean “evacuation time” variation. on a single criterion. This requires the assessment of various
■■ Step 6. Data obtained in the simulation have been criteria and the evaluation of alternatives based on each
incorporated to a new improved multi-criteria decision criterion and then the aggregation of these evaluations to
problem, obtaining the same result as in the subjective achieve the relative ranking of the alternatives with respect
multi-criteria. The NASA alternative is still the best rated to the problem. The problem is further compounded when
and sensitivity analysis contributes to decision robustness. there are several or more experts whose opinions need to be
The first step consisted of the development of a multicri- incorporated in the decision-making. It is actually the lack of
teria opinion project by a team composed of an analyst and adequate quantitative information that leads to dependence
different CIC experts in order to stablish the important and on the intuition, experience and judgement of knowledge-
critical criteria in the CIC layout. able persons called experts.[4]
In the second step, the experts made a subjective judg-
ment of the project and the analyst processed the results by Analytic Hierarchy Process (AHP)
means of an Analytic Hierarchy Process (AHP) systematic The AHP is based on the experience of its developer, T.L. Saaty,
approach. while directing research projects in the US Arms Control and
The third step was intended to provide objectivity to the Disarmament Agency. Saaty noted the lack of an easy to im-
decision making process by performing a simulation of the plement methodology to enable complex decision making. The
CIC personnel evacuation. This required a previous exper- simplicity and power of the AHP has led to its wide diffusion
iment design that showed the influence of all the possible across multiple domains in every part of the world.[4]
layout alternatives in the evacuation time. AHP provides a means of decomposing a problem into
The fourth step consisted of the validation of the simula- a hierarchy of sub-problems which can more easily be
tion results. embraced and subjectively evaluated (Figure 8).Thus, the
In the fifth step, the validated simulation results were problem is divided into a hierarchy of goal, criteria (ci),
added to the multicriteria decision tool. sub-criteria (scij) and alternatives (ai).
of one alternative over another with respect to a common bution of embarked persons, including the possibility that
criterion. Several applications of AHP to ship combat systems some injured personnel may be transported by stretchers.
design can be found in the literature.[7] ■■ The size, location, function and risks of individual com-
of Action Lines Decision Aid. be worn or carried (e.g. firefighting outfits, Emergency
Escape Breathing Devices, life-jackets or personal thermal
Design of Experiments (DOE) protection suits.
The statistical design of experiments (DOE) is an efficient Routine escape, evacuation and rescue procedures must
procedure for planning experiments so that the data ob- be established to ensure effective escape, evacuation and
tained can be analyzed to yield valid and objective conclu- rescue performance, except for inspection, maintenance and
sions. DOE allows the user to identify the factors which any training.
type of process depends and to know the way in which these After designing and establishing escape routes, proce-
influence on it.[8] dures and evacuation equipment, it must be demonstrated
Several DOE models can be identified: that the ship can be evacuated in less time than the maxi-
■■ Fixed-effects model: the designer decide specific levels to mum established by the NSC.
be experienced, so the conclusions can only be extrapolat- IMO and NSC guidelines require crew ships evacua-
ed to those levels. tion to be assessed by computational means during the
■■ Random effects model: levels are randomly selected, i.e. design phase to provide information on the duration of
the values of the levels are a random sample of a popula- evacuation, crew flows throughout the ship and to iden-
tion levels. tify and, as far as practicable, eliminate areas of conges-
■■ Balanced model: all samples are of equal size. tion. Regulation specifies a minimum of four passenger
other methods are to be used to determine the margin of TABLE 2. Subjective multicriteria (ADLA)
buoyancy, stability, ship attitude and the ability for persons
onboard to evacuate following extreme threat damage or field around it and the strength is inversely proportional to
extreme flooding. the distance from it.
The model complies with the guidelines for validation
Evacuation computer simulation modelling: proposed by the International Maritime Organization (IMO)
Pathfinder and PyroSim software at MSC / Circ. 1238.
Among the available evacuation simulation programs,
several were considered, and Pathfinder was chosen. This is CIC modelling with Pathfinder software
the software developed by Thunderhead Engineering[10] that The motion environment is defined by a triangular mesh as
has great graphical 2D and 3D capability and enables the shows Figure 9 drawn to match the actual dimensions of a
integration with PyroSim fire modelling software.[11] Both model. This motion mesh can be entered manually or au-
are user graphical tools based on well proved Fire Dynamics tomatically from imported data (i.e. from the PyroSim FDS
Simulator (FDS) and EVAC developed by the National Insti- geometry).
tute of Standards and Technology (NIST) of U.S. Department Walls and other obstructions don’t generate triangle mesh
of Commerce.[12] so occupants cannot move in it. Door exits are located at
Advance evacuation modelling software is based on the mesh edges, as seen in green color in Figure 9.
Social Force Model (SFM) of Prof. Dirk Helbing, where Crew speed on corridors must be established according to
pedestrian are represented as circles with a radius of their IMO or to experiment data. Speed on stairs is reduced as a
shoulder width. Driving force is assigned to define the function of the slope of the proper stairs.
motivation of agents to move towards the destination with a Each occupant is defined by their position within the
desired speed. Social forces make agents interact with each compartment, their profile (height, speed, avatar, etc.), and
other and obstacles that are within their interaction range by their behavior, which considers instruction sequences; i.e.
along their path to the destination with identical priority. occupants may stay in their places for a certain lapse of time
It is considered that each obstacle has a repulsive potential and then proceed to an elevator or a door exit.
uation time.
■■ Knowing the evacuation times difference between CIC
types.
Thus, a full multifactorial two-factor DOE with three lev-
FIGURE 12. Bar chart with weighing reassignment els (32) was designed, as seen in Table 3, nine experiments
for each type of CIC (27 in total) were performed.
Mobility & Accessibility criterion in favor of the CENTER Factors selected were: number of corridors (0, 1, 2) and
alternative. The SIDE alternative has considerably lower number of escape doors (3, 4, 5). The meaning of number of
values than the others. corridors is free spaces between console groups.
Sample size of simulations for each experiment has been
Sensitivity Analysis set at ten. A priori, it is not easy to estimate the sample
NASA-type layout was valued with less than 4 points over size, therefore it is advisable to overestimate it, but given
the CENTER-type, so a sensitivity analysis is needed to the important time consuming required by simulations, a
check whether these results are sufficiently robust or, by statistical analysis was applied to know the number of rep-
contrast, they may be affected by slight changes in the input licates necessary to estimate the population mean within a
data or in the decision maker’s subjectivity. Although expert confidence interval.
decision makers do not usually change valuations, it is
necessary to analyze how their weights impact in the final Computer modelling
model decision. PyroSim software[11] has been used to design and draw
As shown in Figure 11 and Figure 12, experts must different types of CIC with their factor variations. Pathfinder
increase Mobility & Accessibility vs Command & Control
weight in pairwise matrix until difference between them CIC type Factors Levels
is reduced to 44.94% (72.47% - 27.53%) in order to 0
switch decision in favor to CENTER type. Criteria change Corridors 1
is so high and unlikely that proves the robustness of 2
NASA
decision made. 3 ladder, Stbd, ForePs
On the other hand, SIDE alternative needs to change Doors 4 ladder, Stbd, ForePs, Ps
weighting more than 53% in favor of Mobility & 5 ladder, Stbd, ForePs, Ps, ForeStbd
Accessibility criteria. TABLE 3. DOE factor and levels for NASA CIC
software[10] has been used to perform DOE simulations as Therefore, it is considered that normal distribution fits best
seen in Figure 13. the egress times for this simulation. Figure 14 shows lognor-
Pathfinder software is deterministic in terms of results, mal distribution and normal distribution curves.
i.e.: for given variables of the agents and same compartment A hypothesis test was performed on results achieved to
layout, the evacuation time is constant. Therefore, the fol- evaluate if there were significant differences between the
lowing generic behavior was configured to give the needed sample population mean (ANOVA Contrast), proving that the
stochasticity: null hypothesis was not rejected in favor of the alternative
■■ Evacuation speed: uniform distribution [2.04 m/s to 2.17 with a significance level of 0.05) see Table 4. This means
m/s] based on Spanish Navy drills. that there is not statistical evidence that the factors variation
■■ Door choice: uniform distribution [15% to 100%], mea- designed within the DOE simulation produce different mean
sures agent decision level to go to the nearest door. In case population egress times i.e. factors variation do not affect
of a bottleneck (more than 4 individuals per square meter) the egress time.
if the randomness of the distribution assigns 15%, the agent Although the NASA type gives the longest egress time of
will go to another door with a probability of 85%, while if the three analyzed CIC types, the difference is less than one
100% is assigned, the agent will wait at bottleneck. second and therefore can be considered as not significant.
■■ Initial delay: Normal distribution (20, 6) [minimum 10
seconds; maximum 60 seconds]. This variable assigns Validation of the simulation with real experiment
a delay time to each agent before the evacuation be- This validation consisted of verifying that simulation models
gins. It simulates equipment destruction that CIC crew assume an adequate approximation of reality. For that, real
must do before start to evacuate and watertight doors evacuation experiments were performed on two of the CIC
opening time. best evaluated in the multicriteria subjective decision model:
Before each simulation, “Randomize” command is activat- CENTER04 and NASA04.
ed so new values are assigned to variables and stochasticity Experiments were made in the Spanish Naval Academy
is achieved. Results were satisfactory, since there were no with 40 midshipmen (Figure 15). The perimeter dimensions
bottlenecks of more than 10% of the evacuation time in any and the tables’ layout were as close as possible to the CIC
of the 27 simulations, as required by IMO Msc 1238.[13] room dimensions and consoles layout.
Usually, the experiments of evacuation are simulated with
heterogeneous groups (different ages, capacities or move- Error (s)
CIC Egress time (s) 95% confidence level
ment limitations) that contribute to long evacuation times.
For these cases an asymmetric distribution like lognormal is Side 37.05 0.50
the best choice to simulate egress times.[14] In our exper- Center 37.75 0.56
iment, crew knows the ship perfectly and has the correct
NASA 37.86 0.54
training required through regular ship evacuation drills. TABLE 4. Simulated egress time (Pathfinder)
Error (s)
CIC Egress time (s) 95% confidence level
Center 35.95 2.73
NASA 37.70 3.11 FIGURE 16. Global assessments according to the
TABLE 5. Experiments egress times improved multicriteria
For each CIC type, 6 events were done, every midship- occupant’s orientation, which is very close to the reality
men had a different delay time that varied for each event. of a ship. The CIC staff knows the compartment and is
These delay times were generated from a normal distri- perfectly trained from drill evacuations.
bution (20, 6) [min 10; max 60], same way as made with
Pathfinder software. Improved Multicriteria
Table 5 shows the obtained results, e.g. the average of the With the time obtained in the Pathfinder simulation, a new
6 events and the error with 95% confidence level. improved multicriteria was processes by the ADLA software.
A hypothesis test was performed on these results and Subjective experts’ decisions data in the pairwise compar-
those obtained by software simulation, resulting in null ison matrix of the alternatives to the evacuation subcriteria
hypothesis H0 of equal means not rejected. At a significance in the initial model, were replaced by the evacuation times
level of 0.05 there was insufficient statistical evidence to of the three CIC layouts simulated by Pathfinder software.
assume that the mean of experiments was different from the These times were performed using a linear utility func-
mean of computer simulation. tion, in which 100% of utility was determined for evacuation
Results showed there was no bottleneck for more than times below 20 seconds (it is estimated that there would be
10% of the total evacuation time as required by IMO no decrease in that evacuation time), and 0% of utility was
Msc 1238. determined for times greater than 60 seconds (times are
A last experiment called “blind evacuation” was per- based on real data). Due to the high percentage of casualties
formed for each type of CIC layout, simulating the presence that may occur, more than one minute of evacuation time is
of smoke or a possible power failure and the loss of illumi- considered unacceptable.
nation inside the compartment. The same initial wait time The results obtained from an alternatives’ assessment of the
of 10 seconds was set for all occupants. The obtained egress improved multicriteria can be seen in Figure 16, which shows
time was 50.64 sec. for CENTER and 45.88 sec. for NASA. that NASA alternative to be the best evaluated with 44.75%.
From the DOE real experimentation two main conclusions The intermediate calculations made to evaluate the final
can be drawn: alternatives assessment are shown in Table 6.
■■ The simulation model developed in the Pathfinder is vali- The first row shows the criteria with their weights, which
dated as an adequate approximation to the experimental do not vary from the subjective multi-criteria. In columns,
framework of the problem. below criteria, are theirs contributions to the different
■■ The blind evacuation was quite smooth and fast. These alternatives COMMAND & CONTROL has not been modified.
blind experiments were the last to be completed, and On the other hand, the contribution of the MOBILITY & AC-
probably this contributed to the improvement of the CESSIBILITY criterion has varied: NASA alternative has been
Center
41.15% 28.00%
39.50%
NASA
47.40% 26.22%
44.75%
Conclusions
This work has been intended to be an aid to the decision
that the Staff of the Spanish Navy must adopt for the F-110
frigates optimum CIC layout.
Among the layout alternatives, taking into account cri-
FIGURE 18. Improved multicriteria reassignment teria provided by the experts, the best valued layout is the
NASA type.
reinforced reaching a higher contribution, 26.22%, versus Therefore the following results have been achieved for
the 17.22% obtained with the subjective multicriteria. the three objectives:
Objective 1: Application of multicriteria decision-making
Comparative analysis between subjective and technique for choosing the best CIC equipment layout. It has
improved multicriteria been possible to carry out the AHP methodology with the
The NASA layout was valued as the best alternative in both indispensable participation of experts with extensive experi-
cases. Subjective multicriteria valued it 3.4 points above the ence in ships. These experts are also directly responsible for
CENTER alternative, whereas the improved multicriteria the decision.
gave it 6.2 points more. Objective 2: To reduce subjectivity to the designed mul-
The main difference between the two multicriteria mod- ticriteria analysis. It has been possible to perform evacuation
els is the robustness of the decision. Although the subjec- time computer simulation on a statistical DOE and has been
tive multicriteria was quite robust, applying the objectivity validated through real experimentation.
AUTHOR BIOGRAPHIES
GERARDO GONZÁLEZ-CELA, is a researcher at Spanish Navy RAMÓN TOUZA is a teaching assistant at Spanish Naval Academy,
Operational Military Research Cabinet (Gabinete de Investigación Spanish Navy Commander. Master in Advance Data Analysis and
Militar Operativa Armada-GIMO), Spanish Navy Commander, Modeling Building at Complutense University, Madrid, Master
Master in Decision Engineering at Alcalá de Henares University, in Decision Engineering at Rey Juan Carlos University, Madrid.
Madrid. gerarcela@fn.mde.es rtougil@fn.mde.es
ROBERTO BELLAS is a teaching assistant at Spanish Naval RAFAEL CARREÑO is a teaching assistant at Spanish Naval
Academy, researcher in marine fire prevention and evacuation
Academy, researcher in models and techniques for project eval-
system designing and modeling, with 20 years’ experience in
uation under risk, Msc in Industrial Engineering at Vigo University.
marine fire prevention system designing and commissioning, Msc
in Industrial Engineering at Vigo University. rbellas@cud.uvigo.es rafael.carrenno@cud.uvigo.es
REFERENCES
[1] C. A. Whitcomb, “Naval Ship Design Philosophy [8] L. J. Freeman, “9/10/2012-1 Implementing Design and
Implementation,” Nav. Eng. J., vol. 110, no. 1, pp. 49–63, Jan. Analysis of Experiments in the U.S. Department of Defense
1998. Testing Communit,” in European Network for Business and
[2] “CIC Operations Specialist Training Course.” Center for Surface Industrial Statistics, 2012.
Combat Systems (CSCS), Dahlgren, VA, 2010. [9] International Maritime Organization, “SOLAS, The
[3] “IHS Jane’s Navy International - Digital, Online and Print International Convention for the Safety of Life at Sea,
Magazine | IHS Jane’s 360.” [Online]. Available: http://www. Consolidated Edition.” International Maritime Organization,
janes.com/magazines/ihs-janes-navy-international. [Accessed: London, 2014.
12-Mar-2017]. [10] Thunderhead Engineering, “Pathfinder Technical Reference.”
[4] N. Bushan and K. Rai, Strategic Decision Making Applying the New York, p. 180, 2017.
Analytic Hierarchy Process, IX. Springer New York, 204AD. [11] Thunderhead Engineering, “PyroSim User Manual.” New York,
[5] T. L. Saaty, “How to make a decision: the analytic hierarchy USA, 2017.
process,” Eur. J. Oper. Res., vol. 48, no. 1, pp. 9–26, 1990. [12] K. McGrattan, S. Hostikka, R. McDermott, J. Floyd, C.
[6] T. L. Saaty, Decision making for leaders : the analytic hierarchy Weinscheck, and K. Overholt, “FDS Technical Reference Guide.
process for decisions in a complex world. RWS Publications, Volume 1: Technical Reference Guide.” NIST National Institute
1995. of Standards and Technology, Gaithersburg, p. 175, 2016.
[7] J. G. Michaeli, G. Hou, D. Cornett, and A. Nelson, “Application [13] IMO, “IMO MSC/Circ.1238 ‘Guidelines For Evacuation Analyses
of the Analytic Hierarchy Process for Topside Combat System for New and Existing Passenger Ships,’” London, 2007.
Integration onto Surface Combatants.” [14] T. Meyer-König, P. Valanto, and D. Povel, “Implementing
Ship Motion in AENEAS — Model Development and First
Results,” in Pedestrian and Evacuation Dynamics 2005, Berlin,
Heidelberg: Springer Berlin Heidelberg, 2007, pp. 429–441.