and Human Limitations
Stephen Wright MIMarEST
November 6, 2015
11:45 - 12:15 EST
The presentation today explores the limits and challenges of
remote and autonomous commercial ship operations, especially
regarding US-flag vessels.
We will address the history of commercial vessel automation and
look to its future.
We will explain several key elements that must be quantified and
addressed by a well-designed marine automation system –
especially an autonomous system.
We will help define the line between engineering and popular
PLC ship system
Washington State Ferry control
cabinet. This vessel uses an
engine order telegraph and
electric propulsion – a mix of old
Our objectives today
• Existing commercial ship automation
• Where did shipboard automation come from?
• How fast is it changing?
• What limits its growth?
• How automatic are ships today?
• Remote and autonomous commercial ships
• Are they possible technically? Are they possible socially?
• Is ROV/AUV technology scalable?
• Are present rules sufficient for the ship of the future?
Some points to look for
• What kind of commercial ships are likely to automate further?
• Is the autonomous ship really unmanned?
• How limiting is the lack of technology transfer between the
ROV/AUV community and surface craft designers?
• What if the first automated vessels in the US are built to
American Boat and Yacht Council (ABYC) standards?
• Engine room automation
• Power management and automated systems such as ballast water,
control air, anti-heeling, fuel purification, bilge treatment, etc.
• ACC and ACCU from ABS; similar notations from other class societies
• Bridge electronics
• Integrated bridge systems prove the whole greater than the sum of its
parts. ECDIS and ARPA; solid-state gyro-compasses with
• Auto-pilot systems with memory, like a pilot’s
• Safety systems
• Fire systems with graphic interfaces and automated responses
• Intrusion and perimeter control; auto-ballasting systems; CCTV
• Propulsion systems
• Twin-screw synchronization systems - old school
• Hybrid propulsion
• Batteries and “power take-off/power take-in” systems (PTO/PTI)
• Slow speed marine engine dynamic optimization is now available from
large engine manufacturers to optimize for fuel cost or for schedule.
• Water and cargo systems
• Auto-ballasting; anti-heel systems; automated tank-washing;
automated loading and discharge systems; auto-loading bulkers
A matter of degree
Ships already have centralized
lineups of switchgear actuated
Each of these motor controllers
has a “Hand/Off/Auto” or
It is only a question of how
remote or how automatic.
Automation possibilities - Remote
• Complete remote operation is possible
• Transas and Kongsberg training simulators resolved many issues
• ROV/AUV developments are largely scalable to commercial vessels
• Department of Defense drone deployments are more challenging than
operating a ship at 12 knots.
• Refining oil or building a car on an assembly line are each more complicated
than operating a ship at sea.
• Remote operation is limited by telecommunications reliability and
bandwidth. In short – weather.
STD and VME Busses allowed
plug-in modules, unlike this
more typical shipboard
installation. This board would
take longer to repair and have
a greater chance of repair
error than plug in board.
Automation possibilities – Autonomous
• Completely autonomous operation is possible
• Early programmable logic controllers (PLCs) did what relay-logic and bread-
board op-amp controls could not do. They were special because they were
modular and reprogrammable; expandable and scalable. PLCs automated
discrete activities well and automated processes less-well.
• Process control took PLC batch control to real-time. Continuous process
control streamlined commodity production but in so doing made control of all
real-time processes possible. Still expandable and scalable, the software
possibilities expanded beyond Relay Ladder Logic (RLL) with ASCII subroutines
to C++ and more advanced languages capable of better utilizing processor
power in real time. Batch processing of oil refineries made autonomous ships
Ships like this have push-button
redundancy, often automatic
lead/lag or master/slave
redundant motors or pumps.
Pushing the selector switch can
be automated – but can
changing the motor or cleaning
Automation has been around
• Completely autonomous operation has been possible for a long time
• STD Bus was 8-bit
• VME Bus opened up 16-bit process control using 6800 Motorola processors
• VME expanded to 32-bit and 64-bit versions
• These industrial bus standards, later becoming IEC and DIN standards,
facilitated the equivalent of the “internet of things” on an industrial scale in
• The heart of the Apple, the heart of the MAC – the 6800, was now ready for
service in producing other things, not “merely” processing data.
• But how do things become represented as data?
8088 and 8086-based modules
ran many European ships in
the 1980s. EEPROM chips
were burned with the program
– all 16K of it.
We can still program those
chips in assembly language
and burn them in our PROM-
burner. Around 2000, we had
to buy a gross of the memory
chips to replace one chip.
Anyone need 143 16K 8-bit
Input/output modules –
The five senses of data
• Unlike humans, data has many more senses: pH, salinity, specific
gravity, viscosity, x-ray vision, sonar-sensing, thermal imaging. The
machine is only limited by the I/O.
• Even the most primitive ACCU system aboard ship can monitor any
and every characteristic in 24 milliseconds. Even midshipmen do not
move that fast.
• Analogue to digital converters allow proportional integral derivative
(PID) algorithms to run in stable software instead of as thermally-
sensitive capacitive shorting circuits used in the 1960s.
• Input/Output modules use more than five senses.
This is a marine-rated analogue
input/output device that would
replace six card slots in one of
the previous slides. These are
more reliable that the old ones
and may last the life of the ship.
Input/output modules –
The “five senses” of data
• Each subsystem is a self-sufficient loop doing its part in a larger loop
and answering to the demands of the central processing unit.
Distributed I/O systems use a pyramid structure to keep most of the
processing local and the supervision not burdened with massive data
• This hierarchy of interdependent self-reliance is the key to a ship without a
• Existing ACCU standards will have to be expanded for autonomous ships.
• Required redundancy: N+1 becomes N+2 or N+3 on a ship without a crew
• The technology is all here, the specification has not been completed
Removable programming unit
on the left side of the photo
that replaces a PROM burner
in a modern ship.
Touch screen to the right
replaces a wall of
annunciators and ten-turn
The 1980s ship was
almost as automated
as the 2015 ship.
The technology has improved
but not the culture – at least
not in the US. This three-panel
lineup would be a half panel
today but the ship probably
has the same number of
points monitored...maybe a
The learning-capable automation system
• An automation system can apply simultaneous analysis and
comparisons in real time, learning from system history to better
anticipate responses providing more appropriate system corrections
with each iteration of its ever-improving response curves.
• In an autonomous ship, the system learns the ship just as a crew would, but
all system information is shared, not subjectively compartmentalized, as with
a human crew.
• The engineering challenge is to parse and save the data while gleaning all that
can be learned from it. A complex system has large data needs. There is no
data center at sea.
• What is done at sea and what is done on land is part of the developing
methods of control.
The price of
One benefit of rapid response
to change is reduced failure of
equipment – problems are
The inconvenient side of this
responsiveness is nuisance
How do you evaluate nuisance
trips without a crew?
• The limitations on autonomous vessels are not technical; they are
social. We can build and operate a remote-controlled or autonomous
vessel today. But our neighbors may not let us.
• Only scientific risk-analysis can determine actual risk
• Perceived risk is often at odds with science
• Here it is a relative risk, not an absolute risk. We compare an autonomous
vessel to a crewed vessel and compare the cargo risk and vessel risk.
• The actual risks include equipment failure and malicious interference –
hackers on line or pirates on speedboats.
• The limitations on autonomous vessels are social. Anticipated
skeptics include labor unions and environmental organizations.
• The likely cargoes are water, crude oil, coal, iron, bauxite, Portland
cement, pet coke or other low-cost, heavy cargoes with no time
• The ideal autonomous ship will slow-bell across the earth with
steadfast determination and maximum efficiency.
• The first autonomous ships will not be box ships with frequent stops
and complex loading procedures.
• Tankers and bulkers natural auto-ships, especially in the Pacific Ocean.
Questions for further study
1. Marine Automation Parity:
a. Why does US marine automation trail behind US industrial automation technology by
b. Why does the US trail Europe in marine automation?
c. What role does Navy procurement play in this disparity?
2. Marine Automation Market Penetration:
1. Why is the marine business among the least automated in the US?
3. Marine Automation
1. Are the decision-makers too old?
2. Is it a closed market due to the Jones Act?
3. Do the consumers, rate-payers and taxpayers pay the price?
4. What can be done to help the marine industry catch up to manufacturing
sector in use of technology? What can SNAME do?
SNAME NY Met Section
IMarEST Eastern USA Branch