There’s a rising demand… but Stevie Knight asks, how do we overcome the challenges of tying together existing technologies with data-centric innovation?

“There is firstly an exponential increase in processing power of devices, secondly an exponential increase in communications speed between devices and third, an exponential increase in data storage capacity – so there are big expectations,” says Stefan Goranov of Winterthur Gas & Diesel (WinGD).

His colleague, Wolfgang Östreicher outlines the issues further: “Traditionally, we deliver a very reliable system that allows you to run a low-speed two-stroke engine, 24/7 over an 8,000nm route with full redundancy. But today, we see more and more requests for digital connectivity, data processing… and there’s more to come.”

There are inevitable issues around the practical aspect of tying together various demands, especially since these are not always natural bedfellows. “One of our biggest challenges is to balance scalability and complexity,” says Östreicher. As Goranov adds: “Our issue is, how do we keep the solution simple… and make it future-proof at the same time?”

This is an issue because right now, the industry is seeing waves of innovation washing in on every side. As Espen Kværnstuen of Kongsberg notes, “engineering is moving very, very fast”, Östreicher agrees and explains: “We all have to be prepared to plug in new technologies”.

Part of the answer comes down to sufficient CPU power and modern interfaces. But there’s also a fundamental question about how to arrange the architecture, which Goranov says means breaking the extended structure up into smaller self-contained chunks and interfacing them with other pieces.

Of course, modularity appeals to the platform manufacturers who make a habit of piecing together a range of customisable solutions. As Kongsberg’s Thomas Kristiansen explains “the idea behind our established Autochief system is that it can either be a stand-alone package or used with other control units such as LNG fuel-tank monitoring, engine pressure monitoring and shaft rotation”.

Kværnstuen describes this as “using a common backbone” with adaptors that other, real-time systems from various suppliers can hang their wares. Östreicher adds that for the engine manufacturers, a modular approach has the advantage of allowing various processing elements to be developed and assessed, tested in combination within a larger collection of units, again on a ‘virtual’ ship and then on the real thing. For Kongsberg, “it gives more possibilities as the suppliers can develop their products independently,” says Kværnstuen.

However, this in itself raises a few more challenges: “Often, different vendors have their own solutions, while at the same time, the owner wants a common system.” This is where Östreicher, Goranov and Kværnstuen might end up on different sides of the table: “Everyone wants to retain as great a scope as possible, but that’s not easy. For example, we want to integrate the generators but understandably the manufacturer wants to keep control of them – after all, platform specialists like us can’t answer every request as efficiently as they can.”

He adds the disparity in approach comes right down to the language the machinery speaks: “There’s always a need to find a communication protocol that fits everyone, but there are a lot of them, and every vendor has their preferences. As a result, there’s a lot of interfacing going on… but someone has to adapt and also bear the cost of that adaption.”

There’s a further issue. “So many of those technologies now trending are not that mature,” says Goranov, at least in comparison with more traditional solutions.

As Kværnstuen points out, no matter what’s onboard “you need to shield safety critical elements like machinery, ballast control and power management… as without them, the vessel is dead in the water”.

Here, the gatekeepers come into their own – and they do exactly what it says on the tin (as well as in the movies). For example, WinGD’s dedicated Gateway Unit sits between the internal and external communications, protecting the inner sanctum of the engine control area. “It provides universal connectivity and sufficient computing power and resources to acquire, buffer and pre-process fast sampled signals from the control system,” says Goranov. It works in concert with a higher-level system - WiDE – that can sense and provide early warnings for anomalies and failures, suggesting appropriate corrective actions says Östreicher.

It can get even trickier when dealing with any overarching platform, especially as companies like Kongsberg and ABB are facing raised demands for overall integration: this entails multiple layers and sub-layers which are both linked by buses and gateway units, right down to the sensor level.

However, there’s a tension between connectivity and efficacy.

Marcus Högblom of ABB explains that footprint, inefficiency and build costs also rise hand-in-hand with complexity. Trying to mesh too many of the various communication protocols, fieldbuses and sensors won’t allow for the requisite segmentation that underpins modularity.

More, as Kværnstuen points out that at least at present “these connections are mostly wired… and there’s a great deal of cabling – kilometres in fact - on a modern vessel”. He adds: “While some industrial wireless protocols are coming, it’s not widely used on vessels – a steel hull isn’t the best environment for wireless anyway, the sensors need power anyway - and for safety-critical systems, it’s just not robust enough yet.”

So, on both the physical as well as the system level, integrating new elements into an existing system always demands a conservative, stratified approach along with “a clear hierarchy to the logic” says Östreicher. Högblom puts it more colourfully: “Do it badly and you end up with a real dog’s breakfast.”


While similar challenges are repeated across the industry, some segments have seen the issues soar.

“A few years ago you had a couple of thousand different inputs on cruise ships that had to be controlled – now it’s tens of thousands,” says Högblom. “The automation covers everything from propulsion to aircon, and today it’s like running a small city.”

Still, ABB’s Ability 800xA has a head start: “We’ve already supplied around 12,000 of these platforms, from urban power plants to steel mills,” he says. On a pragmatic note, since it has its roots in these other industries “the owners know the systems will be supported for a long time, an advantage for cruise ships which usually have a lifetime of 25 years plus”, he points out.

It’s just as well that these platforms are the result of cross-sector development. The cruise industry is becoming heir to green technologies that by and large didn’t originate in the maritime world – and what’s more, these are being coupled together in novel ways: “We are running fast into very mixed systems,” says Högblom. “Now there are fuel cells, energy storage systems, Flettner rotors and all sorts coming onboard.”

The new demands are not likely to ease in the foreseeable future. “All the different interests are trying to create their own pathway, and new technologies are constantly coming in but that’s not being matched by older systems exiting; after all, previous generations of ships are still working well on existing protocols, so there’s no benefit to them in upgrading,” says Kværnstuen.

Högblom adds: “From the automation point of view one key challenge is to tackle the increasing data flow for ship operators, which underlines the importance of user-centric designs for efficient vessel operations.

Kværnstuen concludes: “As a result, we will probably see more and more complexity in future… and of course, we are only seeing the start: we are looking at autonomous ships now and that will require even more automation.”


The present reality for most vessels is that “the diversity of equipment makers, data sources and a lack of standardisation” make it difficult to see, let alone evaluate what’s going on says METIS CEO, Mike Konstantinidis. It’s a task not made any easier by the sensor feed, which can be distinctly patchy.

However, it is possible to get the numerous readings to work with, instead of against each other.

The data streams are compared, so that a misfired GPS signal, for example, is seen in the context of the whole picture, allowing anomalies to be ironed out.

Likewise, these new systems can make better use of existing information: calculating trim by subtracting the draft gauge measurements fore and aft doesn’t usually work as the time stamp doesn’t match, “but this changes if you can cleverly interpolate synchronised data”, he explains.

It can also be used to fill in the gaps. Take a simple, but fairly central case - the torque meter that returns the shaft power information: “The sceptical will often point out that these are unreliable, often uncalibrated or just busted,” he says. Despite this, he explains: “If you take the rpm from the main engine, the measurements from the fuel pump, the turbocharger and the scavenge pressure, you can still make a good estimation of the shaft power.”

Still, this last attribute is best supported, not by algorithms, but by a specialised branch of AI - machine learning.

Importantly, rather than founder on complexity, it thrives on it. “The larger and more varied the amount of data feeding these machine learning systems absorb, the more accurate become the results,” explains Konstantinidis. “It differs from the usual mathematical approach because it is not built by formulas and parameters on an existing base… but is constructed by using the data itself.”


Even though many believe that we are looking at a rising number of analytic bolt-ons located in the cloud, unfiltered data is just too big, too messy, and far too expensive to transmit via satellite. As Goranov says: “Connectivity per se is not the main challenge; securely and efficiently transmitting good quality operational data is more demanding.”

Edge computing could be the practical answer for those detailed, analytic processes that don’t need to be carried out onboard.

Mike Konstantinidis outlines the strategy: “Edge computing means doing the filtering, processing, synchronisation and generally getting rid of the garbage while the information is still onboard. That means delivering compressed and packed data through the satellite connection.”

So, exactly how much room does this kind of thing take up - both real and in cyberspace?

Konstantinidis explains METIS’ Wireless Intelligent Collectors, or WICs “are small enough to fit into the palm of your hand”. But, they pack a lot in, forming an onboard network able to connect any instrument, digital or analogue. “It’s pretty straightforward: you put one on the bridge, one on the main engine, one by the flow meters and probably one in the cargo hold,” he says, adding that “even the most complicated of ships didn’t take more than a couple of days to fit, and that was while sailing”.

Finally, Konstantinidis says that once the data’s been cleaned up “it’s enormously small – so small in fact that the customers don’t even notice the workload on the telephone bill”.


“A few years ago you had a couple of thousand different inputs on cruise ships that had to be controlled – now it’s tens of thousands.” Marcus Högblom

“A few years ago you had a couple of thousand different inputs on cruise ships that had to be controlled – now it’s tens of thousands.” Marcus Högblom

Metis’ Wireless Intelligent Collectors, or WICs form an onboard network able to connect any instrument, digital or analogue. Image, METIS

Metis’ Wireless Intelligent Collectors, or WICs form an onboard network able to connect any instrument, digital or analogue. Image, METIS

There’s a need to protect the inner sanctum of the engine control area. Image: WinGD

There’s a need to protect the inner sanctum of the engine control area. Image: WinGD