International Journal

on Marine Navigation

and Safety of Sea Transportation

Volume 2

Number 3

September 2008

245

Combination of Navigational and VDRbased

Information to Enhance Alert Management

M. Baldauf & K. Benedict

Hochschule Wismar, University of Technology, Business and Design, Department of

Maritime Studies Warnemünde, Germany

E. Wilske & P. Grundevik

SSPA Gothenburg, Sweden

J.O. Klepsvik

Kongsberg Seatex, Trondheim, Norway

ABSTRACT: Within this paper results of ongoing investigations will be presented. Main subject of studies is

laid on the present situation of alert management onboard ships navigational bridges and potential use of data

recorded with mandatory ship-borne VDR equipment during normal ship operation to support the process of

on board collision avoidance. The investigations and results discussed and presented in the paper are gained

within trhe work in two different projects on research and technical development. The first is the European

MARNIS – project on Maritime Navigation and Information Services. It is funded by the European

Commission, Department for Energy and Transport. Secondly some of the results presented here are part of

investigations performed under the national RD project "Maritime Safety Assistance Rostock" which is

funded by the German Ministry of Education and Research Berlin.

1 INTRODUCTION

Safe navigation, including collision and grounding

avoidance, is the overall task of the navigating

officer in charge. The global aim is to ensure the safe

transport of goods during a ship's voyage from port

of departure to port of destination. Modern ship

bridges are highly-automated man-machine systems.

Safety and efficiency of the ship operations are

dependent, as in all other complex man-machine

systems, on the communication between humans and

machines during the accomplishment of tasks.

Humans can fulfil their assigned monitoring, control,

and decision tasks most effectively, if the

information flow between them and machines is

adapted to the human skills and abilities (e.g.

Brainbridge (1983) and Lützhöft, 2004).

In the last years a strong increase of modern

information systems on the ship bridges could be

observed. Simple displays and control systems were

supplemented or replaced by complex computer-

based information systems. Information of different

sensors and systems are combined in integrated

navigation systems (INS). In order to support the

mariner effectively on board, a task- and situation-

dependent presentation of the information is a

compellingly need.

With the enlarged number of systems and sensors

onboard, and the increase of automation a

proliferation of alarm signals on the bridge is

associated. Alarm signals coming from various

systems and sensors sometimes lead to a confusing

and difficult manageable situation for the mariner,

which is distracting him from his task to safely

navigate the vessel. Redundant and superfluous

audible and visual alarm announcements are

appearing on the bridge.

The International Maritime Organization (IMO)

IMO has recognised this situation and decided to

revise the exiting standards for INS and to develop

requirements for an alarm management system. A

working group coordinated by Germany was

246

established to progress this work. Further activities

as especially the eNavigation (Earthy, 2006a)

initiative is strongly linked to thise subject and

covers problems related to integration of systems.

Several investigations were performed to analyse

the specific situation on board ships regarding the

occurrence of alarms by means of a series of

empirical studies. Basing on the gained results and

taking into consideration existing drafts for

enhanced alert management a concept was drafted

for the specific task of collision avoidance as a first

approach. The concept covers a technical

combination of different sensor information together

with the use of data recorded by VDR systems to

contribute to the reduction of the high number of

alarms and warnings presently observed on board of

ships.

The investigations were partly performed under

the framework of a national Research and

Development project funded by the German Ministry

of Transport Building and Urban Affairs, and under

the European MarNIS – project, funded by the

European Commission, Department for Energy and

Transport (Willems & Glansdorp, 2006).

2 PRESENT SITUATION

2.1 Integrated Navigation Systems

According to the existing IMO Performance

Standards a integrated navigation systems (INS) is

defined as a system that supports safety of

navigation by evaluating inputs from several

independent and different sensors, combining them

to provide information giving timely warnings of

potential dangers and degradation of integrity of this

information. By now three different categories A, B

and C of INS are established. The lowest level INS

category A has as a minimum to provide the

information of position, speed, heading and time,

each clearly marked with an indication of integrity.

The second level category INS (B), is defined as a

system that automatically, continually and

graphically indicates the ship’s position, speed and

heading and, where available, depth in relation to the

planned route as well as to known and detected

hazards. Finally INS(C), is a system that provides

means to automatically control heading, track or

speed and monitor the performance and status of

these controls.

The definitions and categories are under

reviewing presently. One aim of the work of the

IMO Correspondence Group reviewing the standard

is, to elaborate a more generic definition and start

with a totally new functional approach, where tasks

and functions will use as the basis for INS definition.

In future INS will be defined as such, if it performs

at least two of the navigational tasks route

monitoring, collision avoidance and track control. Of

course further navigational tasks may also be

integrated to such systems. Additionally an alert

management system should become a mandatory

part of a future INS and is specified in the draft

standard in a separate chapter.

Table 1. Required mandatory IMO alarms for selected

navigational devices

__________________________________________________

INS component / Number of mandatory alarms

___________________________

Device

__________________________________________________

Heading Control System 3

Track Control System 10

ECDIS 7

Radar 5

GNSS 7

Echo sounder 2

Gyro compass 3

Bridge watch 1

alarm 2

AIS 6

VDR 2

__________________________________________________

A inventory looking at the required navigational

alerts was performed and the result is summarized in

Table 1.

2.2 Analysis of alarm management

In the frame of different projects a series of several

empirical studies were performed to analyse the

present state. These studies were aimed at the

improvement of ship borne alarm management of

INS. It was started with a review of kind and types

of alarm messages implemented to navigations

systems. Considering the technical integration of the

relative new Automatic Identification System (AIS)

the situation is summarized in the following figure.

It can be seen, that the number of implemented

alarm messages to increases rapidly with the

interconnection to other navigation systems and the

level of integration consequently.

Moreover the example shows, that the number of

real alarm messages implemented to the systems is

much more higher than the number of required

mandatory IMO alarms. The number of potential

alarms and the design of alarm systems in general

seems to be a problem (Earthy, 2006b).

247

Fig. 1. Number of technical alarm messages implemented to

selected stand alone and connected systems

Further for this purpose real alarm situations on

board of vessels were continuously recorded (e.g.

Lepsoe & Eide, 2005). Interviews and questionnaires

with experts were used additionally to collect data

about operational needs of the navigators (Motz &

Baldauf, 2007).

With respect to kind and frequency of alarms a

great variety was detected. A discussion of the

gained results showed that strong correlations are

indicated in respect to the area related navigational

situations (open sea, coastal area and confined

waters). As one of the main reasons it was found,

that the implemented alarm algorithms are fixed, as,

e.g., collision warnings, having no suitable technical

option to adapt automatically alarming to changing

conditions of a navigational situation. Figure 2

illustrates this result exemplarily.

3,2

10,8

26,2

0,0

5,0

10,0

15,0

20,0

25,0

30,0

open sea coastal confined

Navigational situation

Frequency of alarms per hour

Fig. 2. Average frequency of alarms depending on sea areas for

a voyage of a passenger vessel

On the other hand all the studies have also shown

a great dependency of the alarm rates on the specific

navigation devices integrated on the bridges. As can

be seen from the example given in Figure 3 the

majority of alarms were registered at the radar

device and were dangerous target alarms mainly.

132

100

120

140

Heading control

Track control

Radar

ECDIS

GNSS

Echo Sounder

Gyro Compass

Alarm. Managem. engine.

Bridge alarm system

Fire alarm system

Ballast water display

Equipment

Frequency of alarms

Fig. 3. Frequency of alarms per device

Basing on the results it has been concluded, that a

future alarm management should harmonize the

operation, handling, distribution and presentation of

alerts. To avoid the uncontrolled increase of alarms,

a set of priorities based on urgency of the required

response is needed to improve the operator’s

situation awareness and his ability to take effective

action. Therefore a new philosophy is suggested for

the prioritization and categorization of alarms. Alert

is defined as umbrella term for the indication of any

abnormal situation with three different priorities of

alerts:

− Alarm (highest priority) - immediate awareness

and action required;

− warning - awareness of changed condition;

− caution - awareness of condition which does not

warrant an alarm or warning condition, but still

requires attention out of the ordinary

consideration of the situation or of given

information.

The three priorities should be indicated visual and

acoustically in different ways.

To categorize the alerts further, the following two

alert categories are specified at the moment.

− navigational alerts - functional indication of

dangerous situation, e.g., collision warning, depth

warning;

− technical alerts - equipment failure or loss.

Basic concepts for improvement of collision

warnings are already available but has to be further

researched and developed. Investigations should be

dedicated to apply the concept according to a

functional approach for new alarm management.

Finally, the alert management HMI should be

integrated to support the bridge team in the

immediate identification of any abnormal situation,

including the source and reason for the abnormal

situation and in its decisions for the necessary

248

actions. The alert management HMI should be

provided at least at the position from where the

vessel is navigated and fulfil two major functions:

indicating and identifying alerts, allowing the

acknowledgment of alerts by the bridge team.

2.3 Case study of a real collision scenario

A couple of real situations were investigated to

analyse the present situation regarding the

operational integration of AIS for the purposes of

collision avoidance and to identify potential existing

technical lacks and problems as a basis for better

integration of this new technology into the

navigation process on board. One example of a

collision scenario is used her to highlight some

aspects of integration.

The tracks depicted in Figure 4 belongs to an

example of a scenario happened in open sea area

with a traffic separation scheme during night time

and conditions of good to moderate visibility. Four

vessels (A-D) equipped with AIS were involved in

the situation that led to a collision between target

"B" and "D". These tracks were produced by using

an ECDIS based software on basis of AIS data

recorded with own AIS equipment installed in

laboratory for scientific investigations and operating

in the "listen-mode" only.

Fig. 4. Target tracks reconstructed on basis of recorded AIS

data

For the purposes of a more detailed statistical

analysis a section of 20 minutes was taken from the

continuous record. The data were modified and

made anonymous before the analysis was started.

The analysis of the available AIS data was

performed with a self created software and numerous

results were gained. The intention of the analysis

was not to identify who is to blame but to highlight

the operation of AIS for further integration of the

system into the process.

The main outcome of this analysis can be

summarised as follows:

Regarding voyage related data:

− The destination data of at least one of the

involved target were not correct, because course

data do not correspond with port of destination

− The used format for port of destination and ETA

did not comply with the recommendations of the

IMO,

Regarding the dynamic data:

− The heading information of one of the involved

vessels was obviously wrong (great difference

between COG and heading),

− The optional AIS information about rate of turn

was not available for major part of the

observation period

− Regarding the transmission scheme:

− The standards defined by the responsible

organisations require shorter update rates and

transmissions if the ship is manoeuvring.

Although several course changes were registered

from the record, no change in the update rates of

the relevant targets was observed during the

whole tracking period.

− Furthermore, a number of duplicated and tripled

messages, probably from repeater stations, were

registered.

On one hand the results of this spotlight confirm

the known problems about the reliability respectively

the uncertainty of AIS data. On the other hand it also

emphasises the necessity of great carefulness when

using data of the system for decision making for

collision avoidance. The most important issue that is

illustrated by the given example is that AIS should

never be used as the sole source for decision making.

However, with respect to the potential of

integrated navigation systems the example shows

also, that there is a urgent need for enhanced alarm

management. Obviously no alarm has influenced the

situation awareness of the responsible watch officers

as intended by the IMO rules. From whatever reason

of the officers behaviour that lead to the collision –

maybe the dangerous target alarm was not switched

on at all - it is to simple, to categorise the cause of

the accident as a so called very popular "Human

Factor". If that is recognised, then it has at least to be

taken into account, that there were more than one

"human failure". Regarding the alarming it has to be

concluded again, that alarm algorithms has to be

much more enhanced. More effective alarms are

needed in a way that the only occur in condition,

when the operator is really oversee critical

conditions and there is a real need to support and

complete the officers' situation awareness. This is

presently not the case, especially for the dangerous

249

target alarm algorithms implemented to AIS,

Radar/ARPA and other integrated systems. That is

on reason, way it is allowed to switch of dangerous

target alarms.

3 COMBINATION OF INFORMATION

3.1 Performance Standards on INS

The existing as well as the reviewed INS

performance standards call for combination of

systems, data and information '... to provide

information giving timely warnings of dangerous

situations …'. As already mentioned above, for that

purpose, three different priorities indicating three

different levels of urgency for taking actions are

suggested. This approach may be also useful for the

enhancement of dangerous target alarms.

As investigations have shown (inter alia by

Baldauf 2006) there is an unsatisfactory exchange of

data and information available on a ship's

navigational bridge from different sensors and

sources. Taking into account the behaviour of

navigators it is clear that alarm thresholds for

dangerous targets are depending on situation

parameters, mainly sea area and visibility but also

own ship's and target's speed and course, sizes,

manoeuvring characteristics and so on. However,

even the simple connection of CPA calculation

considering also information on land masses and

fairways available from ECDIS are not in use until

now. Also the change of manoeuvring characteristics

depending on ships speed or loading conditions is

not yet covered by any dangerous target alarm

implemented to INS.

3.2 Approach to reduce "Dangerous Target"

alarms

Enhanced alert management of future INS needs

more combined use of available information also for

triggering dangerous target alarms. For that purpose

a first generic concept is drafted to combine target

information from different sensors and manoeuvring

information that could be triggered by VDR or

alternatively also from ECDIS recordings (see

Fig. 5).

Core element of the approach is a risk model for

situation assessment. It is applied to the IMO's

approach of three priority alerts. It is used as a basis

for approaching to situation dependent triggering of

dangerous target cautions, warnings and alarms as

well. By now it is assumed, that cautions and

warnings may be switched off by an operator,

whereas alarms may not.

The concept of self adaptation of thresholds

contains algorithms respectively options to take into

account and process sea area and visibility

information as well. The manoeuvring information

will then be used for automatic adaptation of the

TCPA related limits of the dangerous target alarms

presently in use.

Fig. 5. Generic concept for use of combined information of INS

to self-adapting and triggering situation dependent dangerous

target alarms

The calculation of the situation dependent

thresholds is considering the real ship dimensions

and the type of encounter situation as well.

3.3 Application of a risk model for situation

assessment

Situation assessment is a fundamental basis for

actions to avoid collisions. The main problem until

today is the lack of missing common parameter and

criteria for situation assessment. Hilgert and Baldauf

have already investigated this problem with regard to

the need for “on board” assessment of encounter

situations taking into consideration the aspect of an

existing or developing risk of collision.

The responsible navigational officers have to

ensure that the risk level given as an expression of

the danger of a collision must be kept on an

acceptable level during the entire voyage. Hilgert &

Baldauf (1996) has developed of a COLREGS based

model consisting of four risk levels. These levels are

derived from the actions required by the relevant

steering and sailing rules. For the “on board” use of

the model variable limit values were defined. They

can be calculated and compared with actual

measurement values of CPA and the actual distance

between ships involved in the encounter situation.

An overview on the model is given in following

table.

250

Table 2. Simplified risk model for situation assessment

risk level

limit values and criteria

Level 1:

risk of

collision is developing

CPA < C

and RNG > R

Level 2:

risk of

collision exists

CPA < C

and R

< RNG ≤ R

Level 3:

danger of

collision is developing

CPA < C

and R

≤ RNG ≤ R

Level 4: danger of

collisions exists

CPA < C

and RNG < R

With respect to the IMO's priorities risk level 2

corresponds to caution (situation requires attention

out of ordinary consideration), risk evel 3

corresponds to warning (changed conditions, which

may become hazardous, if no actions is taken) and

risk level 4 to alarm (immediate action is necessary).

The limit values defined for the model have to be

adjusted according to the type of encounter (head –

on meeting, overtaking or crossing courses),

visibility (good or restricted) and the dimensions of

the ships involved. C

is the limit value for the

minimum safe passing distance which must be

compared with the actual CPA. Dependent on the

specific situation parameters C

varies between 0,25

nm and 1,5 nm. R

, R

and R

are limit values for

distance borders mainly dependent on the relative

velocity of the approaching vessels and response

time limits for necessary actions to comply with the

rules. These values correspond in general to ARPA

limits of TCPA.

With respect to an encounter situation on crossing

courses under conditions of good visibility the

"assessment range" R

represents that distance from

which the "stand on" vessel has to keep her speed

and course. If the “give way” vessel has not taken

appropriate action up to the "manoeuvring range"

, so the "stand on" vessel may take action in

accordance with rule 17 (a) (ii). The last limit value

represents the "critical range" R

which

indicates that limit for the "stand on" vessel where

she has to initiate her evasive manoeuvre according

to rule 17 (b) (see following figure).

Fig. 6. Application of the risk model to a encounter situation on

crossing courses under conditions of good visibility

3.4 Use of data recorded by VDR or ECDIS

Recorded manoeuvring data should be used to adapt

the limits in a INS according to the prevailing

circumstances. Therefore the continuous recordings

of integrated systems should be used and relevant

data streams extracted to either a manoeuvring

database or a simulation module that calculates the

necessary manoeuvring parameter on basis of actual

measurements. In this way the thresholds for ranges

, R

, and especially R

) respectively the

corresponding times (especially the time for a course

change of 90° as basis of R