Motivating effort to improve vessel performance - from People Tech Maritime Athens

Christos Papandreou, CEO & Founder, Blue Autonomy

To achieve vessel performance, first you need to measure performance and decide what to change. Then you need experts who can drive this change, working with the crew and operations staff. Christos Papandreou explained :

Improving vessel performance can be more a motivational challenge than a technical one, said Christos Papandreou, CEO & Founder, Blue Autonomy, and a former fleet performance engineer with Athenian Sea Carriers, speaking at People Tech Maritime Athens in April.

“Our industry is still in its infancy on [knowing] what drives efficiency.”

A typical story is that a company decides it is time to get serious about saving fuel, so they decide to buy a product, and evaluate vendors, run pilots, and buy a state-of-the-art monitoring system. “Real time telemetry, slick dashboards, predictive models, everything you could want.”

After 6 months, data is flowing into the system, but no-one is looking at it. “The alerts go unacknowledged. The system is collecting digital dust.”

The weekly performance reports are sitting unread in someone's inbox along with 3000 other messages.

This happened because the company believed that it would get outcomes just by purchasing the technology. Operations staff were never persuaded to do anything differently.

“If people responsible for day-to-day operations do not realise why these things matter to them personally, no amount of software could move the needle. Technology gives you the ability to see problems but does not give you the willingness to solve them.”

Measuring and deciding

Improving vessel performance starts with measuring performance. The traditional approach, based on ISO 19030, is globally recognised and understood by everyone from paint manufacturers to charterers. Every step is documented and every assumption is visible.

“There's nothing hidden behind any proprietary algorithm,” he said. “It has been the backbone of hull and propeller performance assessment for years. It's straightforward, reproducible, and transparent.”

You still need someone with technical knowledge to determine what to do with the results. It is only measuring performance, not telling you how to manage it.

This has many complexities. For example, an antifouling coating can reduce friction through the water, but it will be engineered for a specific operating profile. If you are going below the recommended minimum speed or have extended stationary periods in warm and shallow coastal waters, you will get more biological growth, and the coating will not perform as expected.

An alternative way to measure performance and make decisions is to use tools with machine learning. These can absorb many different variables and see patterns a human analyst could take months to find.

When deployed properly, they can offer remarkable predictive capability, he said. But you still need a domain expert to validate the predictions.

Mr Papandreou has seen a neural network-based system which predicted that a vessel would use less fuel when going against a headwind than when there was no wind. “The prediction was physically impossible,” he said.

“This happens when you treat the model building as a pure computational exercise, when you skip the step where an experienced marine engineer looks at the feature set and says, ‘wait, that variable does not sit well.’”

“The most powerful approach I have found is to combine algorithm capability with human domain knowledge. Let mathematics find the patterns in the noise, have the professional decide which of the patterns is real and which is fabricated,” he said.

Data infrastructure

The information backbone for operational data is essential, including the collection, validation, and delivery of data. If this data is poor, any conclusion drawn from it will be poor. “Without it, we are navigating blind,” he said.

You need a proper architecture for the data flows, “not a cobbled together spreadsheet passed around by e-mail.”

You need to ensure the sensors are calibrated, reporting at consistent intervals, and sending data automatically wherever possible.

You need a verification stage for the data. “Vessel data is noisy, inconsistent, sometimes flat out wrong,” he said. “You need systematic checks before everything reaches the analyst screen.”

This will need domain expertise. “I have audited fleets where almost half of the operational data was compromised in some way. Gaps, contradictions, impossible readings. No visualisation tool will warn you about this. It takes someone who knows.”

For example, if it is a 21,000-dwt bulk carrier going 12 knots in the South China Sea, a domain expert would know what range the data is likely to be in.

Organisational structure

Then you need the organisational structure to achieve the improvements. For example, when you have worked out that fuel consumption is increasing 8 per cent due to degrading hull condition. You need to convince the superintendent to schedule the maintenance task and convince the fleet manager to change the voyage planning.

You need expert interpretation of the data, from someone who can understand the operational context and turn abstract metrics into concrete procedures. This person needs to have the trust of the staff who would put the actions in place.

You need real commitment from the whole organisation, something more than a memo from the CEO, or a new KPI.

Ultimately, the success of every initiative comes down to people, not process. Sustained expert-led engagement makes the difference between having information and changing behaviour.

The expert can work with the company to convert the system’s analytical findings into step by-step operational guidance, not just reports. This person can design workflows that become part of the daily routine. They can build a financial case so compelling that even the most sceptical stakeholder cannot ignore it.

“Behavioural change doesn't happen in a workshop. It happens between quarters or years,” he said.

You are likely to encounter a lot of resistance. Many maritime decision makers have seen many promises from technology which does not materialise, so the scepticism is rational.

The way to get past it is “small, repeated demonstrations that this works,” he said.

Documentation from these pilot projects should show clear savings. They should show what specifically improved, what went well and what didn't. There should be a clear financial thread running through it.

The ultimate goal is to evolve the organisation to the point where operators make efficient choices automatically, not because a system told them to, he said.

Seafarers

It is essential to have support from seafarers for the vessel performance project. Never give them more administrative work to do. Instead, find ways to lighten their workload with automation, he said.

You should never issue instructions to crew without context. Explain the reasoning and show the benefit. Never assume that technical knowledge is automatically transferred, he said. When someone understands the purpose behind the task, the quality of execution transforms completely.

In one example, Mr Papandreou was involved in deploying a state-of-the-art weather routing optimization system, but the officers totally ignored its recommendations. “They didn't trust a computer to plan the route.”

To try to persuade them, the company invited the crew to do their route planning alongside the models, so they could see how their route compared.

In this case, the evidence to the crew was overwhelming, showing that the optimised route was better. Eventually, crew were demanding the optimised routes before departure. “Nobody forced them to do it.”

In a second example, Mr Papandreou’s team developed a system to advise when to activate a second generator. They worked together with vessel engineers in building the system, to incorporate their judgement about where the threshold should be, based on what the engineer thought the safety margin should be. “We didn't override their expertise; we augmented it with data,” he said.

“Collaborations produce adaptation. Imposition produces resistance.”