Automation failings

22 Aug 2017
Snags: automation is the future, but perhaps not the immediate future

Snags: automation is the future, but perhaps not the immediate future

Automation has yet to deliver its promised productivity. Stevie Knight asks what's the problem

Despite the hype, not all – or perhaps even any - automated terminals are currently delivering according to specifications or expectations.

In fact, DP World’s Jan Cuppens, speaking at TOC Europe, pointed out that productivity currently seems to sit in direct correlation to the amount of automation, which it shouldn’t. Where one semi-automated terminal might make 38 moves per hour, the same terminal under full automation might only make 24 moves an hour.

“That’s productivity like the industry had in the 1990s - when we had single lifts, smaller cranes and nothing was automated,” he says.

While he is “100% sure” that the industry will get there, it needs to take a long, hard and collaborative look at what’s going wrong.

Reliability is a key factor and suppliers need to raise their game to meet the new demand. Terminals also need to dive deep into maintenance. Not only does automated kit have more equipment that can fail - such as scanners, lasers and so on - even a ‘back of the envelope’ calculation shows up a fundamental issue. Given today’s typical Mean Moves Between Failure (MMBF) rate of between 1,200 and 1,400, a vessel being worked by six automated cranes, each responsible for some 240 boxes every eight hours, could expect one of its quay cranes to break down every shift. The resulting manual interventions can kill productivity in an automated terminal.

More, a corresponding fleet of 24 L-AGVs operating at current average failure rates would, statistically, also result in rescuing one L-AGV during the same period.

“We need equipment that is more reliable than the values we typically have for equipment on manual terminals,” says Mr Cuppens. “This is a vendor/user, collective responsibility.”

Operating systems also need attention to be “more stable, providing rolling updates.” In summary, “it’s just not good enough”, he says.

Mixed components

Some automation-related issues have arisen because of the how different components have evolved, says Dennis Kogebohn of Hamburg Port Consulting. Vendor consolidation gives the impression of a one-stop-shop solution, “so you are led to believe you can ask for an automated terminal, pay the bill, then come back in three years and it’ll all work”.

However, those companies initially designed their automated products individually, so when installed, “you soon realise they’ve not been strategically developed to sit together very nicely”.

Martin Mannion of Aecom adds the automation process “can be risky”, with implementation taking longer than scheduled, particularly when it involves a number of new technologies.

Both Mr Mannion and Mr Kogebohn agree it takes quite a bit of work to unify the different elements. But despite the obvious temptation to move the responsibility toward the suppliers by, for example, ordering the ASCs and TOS together, Mr Kogebohn believes a facility needs to do whatever it can to keep integration skills in-house. “It’s the terminal as a whole that needs to be optimised, not just the vendors’ products,” he says.

Crane drivers

Mr Cuppens adds that the human factor – or rather, lack of it – is partly responsible for lowering the efficiency of automated quay cranes. “Taking crane drivers out of the quay cranes could mean that the cycle goes 30 seconds slower.” That could equate to a drop of eight or 10 boxes per hour.

Good crane drivers raise efficiency by working with the inertia of a container, swinging it forward and trollying to bring it to a stop. But the anti-sway stops this from happening in an automated terminal, meaning that the terminal is not using its cranes to their full potential.

There are plenty of checks and balances to keep the automated operation safe enough, so why allow the system to flatten a potential efficiency, he asks. Part of this is down to definition: we currently talk about anti-sway when we should refer to it as “controlled” sway. “We have to start thinking differently,” he says.

However, Mr Kogebohn makes the point that at an up-to-date terminal, it’s largely the skills of the operators that make the difference. “You can always speed things up in a manual terminal much more easily than you can in an automated environment, so that manual operations will almost always outperform automated ones during peak periods.”

Human skills, he says, are hard to replicate by machine: “We have quite a way to go to see a solution which delivers the same productivity as a human.”

Exception handling

Exception handling can also be an issue and terminals are dealing with too many workarounds. Says Mr Cuppens: “In fully-automated terminals there is hardly ever a plan B: if it fails, it fails, and it’s difficult to recover.”

Mr Kogebohn has his own take on the trouble: “In a manual operation, you set it up and anything you haven’t thought through, you just find workarounds for. This won’t succeed on an automated terminal.”

To address this, he recommends rigid planning that captures what he calls the “standard deviations” – reefers, out of gauge cargo and so on – and maps each step of their handling, so there’s as little ad hoc processing as possible. “It means additional effort to run through the exercise at the project stage,” he admits, but the payback is “fewer surprises”.

However, in his view, non-standard cargo could deliver an advantage. Mr Kogebohn points out that at present the balance between standardised handling solutions and the necessary customisation for any automated terminal is usually in the 80:20 range and much of it will be down to “local specifics” such as Australia’s nationwide food control checks.

So, even if the vendors – pushed by the big operators - manage to offer more standardised solutions and interfaces, is there any value in being the same as your neighbour? “Don’t get me wrong, there are synergies to be gained from standardisation, but if you can tweak your system you’ll make the most of your competitive edge.” Added to which, it’ll be harder for others to copy.

To the test

Mr Kogebohn argues for more upfront work in the way of stress testing the TOS. It isn't enough to run any old simulation as some oversimplify what gets thrown at the system. For example, "you can either import a stochastically derived rate to cover unproductive moves... or have something that remembers that a given container is under three others,” he says.

Melbourne, for example, “really wanted to understand truck turn times, and an average wouldn’t do as they needed to know how much shuffling could be necessary – to the 95th percentile – before the truck arrived”. He adds: “If the port authority ties penalties or incentives to waiting times, you’d better know what’s coming."

Added to the mix is the next layer of automation with autonomous trucks. And, artificial intelligence is beginning to pick up speed, which means that equipment may soon begin to imitate experienced drivers. Further, the TOS itself could start embedding machine learning capabilities allowing exception handling to be learnt by machines.

However, while undoubtedly intriguing and novel, these concepts won’t pick up all the pieces. Suppliers still need to get around the table and come up with solutions that will make their offerings more robust. Yet, commentators in the industry remain hopeful that within a year or so substantial improvement in automated productivity can be achieved with vendor cooperation.