Improving vessel performance data

We can improve performance data with time and effort, having standardised formats, reducing sensor errors, supporting crew, and with automated monitoring. Speakers from Angelakos Hellas, Central Mare, Blue Autonomy and NeptuneZero discussed

Companies which win the decarbonisation race will be those which put most time and effort into vessel performance data, 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.

The winners are not necessarily the ones which spend the most on technology, he said.

 Just as with football, budget is an important factor, but not the only one.

Fixing problems with vessel performance data can be a lot of trouble, said ​Stathis Kalkounis, Energy & Environmental Engineer, Central Mare.

Shipping companies are required to send data to a data validation service, which will identify errors or missing data. Then the company has to go back to the crew to ask them to fix it. They may find that the vessel is no longer in the company’s management.

When data is inconsistent between the company’s own platform and the data verifier’s platform, you have to go through it all line by line, he said.

Charterers want to see performance data in the format required by their own software and data verifier, and “their way of thinking”.

If you have poor data, you can end up in a discussion with the charterer about when exactly the vessel came off hire, he said.

Draught sensor problems

Failures with draught gauges can be common, and it may be better to have a backup gauge, said Ioannis Makris, Technical Manager, Angelakos Hellas, moderating the session.

The draught sensor may be the most problematic sensor on vessels, agreed Antonios Lalechos, CEO & Founder, NeptuneZero, formerly projects and performance manager with TMS Cardiff Gas.

But draughts do not change every day, it should be possible to spot a failed sensor if there is a sudden jump in the reading, he said. The draught normally changes very slowly during a long voyage, as the vessel consumes bunkers and water.

Draught sensors can deform over time, which changes their accuracy. They are also very easy to rupture, he said. Shipping companies expect them to last for the lifetime of the vessel, but they would be better treated as a consumable item.

It is also common that crew report draughts incorrectly, even when the sensor is working.

Crew are required to provide a correct draught reading every time the vessel enters and exits a port, and it is checked by port authorities, he said. You can also ask outside companies to measure the draft.

Identifying sensor problems

There have been problems with flowmeters when changing over fuels, Mr Makris said. “We’re not at the accuracy level we want.”

Today’s systems can identify when a fuel changeover is happening, Mr Lalechos replied, including from a change in fuel temperature and fuel flow rate, or fuels flowing in different places. So you can identify a problem with flowmeters happening at a fuel changeover.

Overall, the problem is generally more with how sensors are managed than the sensors themselves. “The quality of [most] sensors is much higher than the industry needs,” he said.

It is important to regularly check sensors and do regular maintenance, as you do  with all other equipment.  Companies often assume sensors are “fit and forget”.  

Gathering reliable sensor data is something fairly new for the maritime industry, and it does not have many skilled technicians.  But it is easier now than it was when the first sensors were introduced, about 9 years ago.

You don’t necessarily need the same sensor service provider for all of your fleet, or the same system on all your vessels. But you do need a system to analyse the data and check it, and someone accountable for making sure it is done, Mr Lalechos said.

With high frequency data, you need at least enough to cross check one set of data against another. If you are not cross-validating data all the time, it will start to drift or develop errors, and you will not be able to rely on it.

You can use data to build simulation models. These will make it clear where data seems to be wrong, and help you fill in missing data. It may identify, for example, that the vessel should have 60 tonnes more fuel in the tanks than the crew are reporting.

There is still much manual work involved.  “We haven’t yet built up manpower inside the companies that are able to cope with the workload,” he said.

Tank level sensors can also be problematic and hard to calibrate. It would be better to replace them once a year. These sensors can cost as little as $100.

Cruise ships do not usually have problems with sensors, because they are willing to spend more money on operational costs, he said.

With some sensors, only the manufacturer is allowed to calibrate them, and this raises the costs, Mr Makris said.

“I think we need a performance officer on board who will be specialized to do calibration of the sensors,” said Ioannis Makris of Angelakos Hellas. “It’s not a matter of training.

But much of the sensor monitoring can be done by a remote performance officer, who can check the data, and tell the crew they need to check a sensor which appears to be failing, Mr Lalechos pointed out.

Crew

Well trained crewmembers can support the performance efforts, said Mr Kalkounis of Central Mare. Training can involve office staff explaining the importance of the procedures to capture and manage performance data correctly every day.

In the training, office staff can explain how the company needs to be able to report correct data at any time to outside companies, and there are costs associated with errors or omissions.

Company staff can come onboard and do an energy audit, sitting together with the chief officer and captain onboard.

The focus on performance can tend to fade away during the voyage. “We need to  remind them every time we have a chance how important this is for the office and  reporting requirements to the charterer,” he said.

Is data getting better?

Today’s data quality is completely different to 10 years ago, said Mr Lalechos of NeptuneZero. Company requirements for data quality have changed, the required transparency of information has changed.

10 years ago, data was entered in noon reports to comply with charter party requirements.  People had a lot of trouble trying to make performance reports from this.

In one company operating over 100 vessels, neither technical nor operational staff were looking at fuel consumption data. And the company did its own chartering, so there were no claims from charterers. Because no-one gained any value from the data, the data was allowed to get bad.

Now performance data is gathered to meet regulations, and needs to be validated and audited.

The data also improved through the work of vessel performance managers. “I used to do this job.  I saw the difference when we started looking at data,” Mr Lalechos said.

“I totally believe  we now have much better data,” agreed Mr Papandreou from Blue Autonomy. “10 years ago, everything was a bit chaotic.”

Mr Kalkounis from Cental Mare also agreed data is much better than it used to be. “The granularity of the data now is much better, it's more accurate.  Regulations have been a trigger for that.”

Data standardisation

It would be much easier to manage performance data if it was gathered and stored in standard formats, said Mr Papandreou of Blue Autonomy.

Vessel data is generated by many different vendors, and comes in different formats. That goes for both high frequency data and noon report data. And we currently use multiple systems to translate data from one format and system to another, and companies store the data in different ways.

Regulators are making a big effort to push for standardised data formats, but the scope is limited to data required under regulations.

It would be helpful if noon reporting had a unified form, rather than each company having its own format, Mr Kalkounis of Central Mare agreed. The varied formats makes it much harder to put the data together. Company staff have to spend time “building bridges” between systems.

There’s one group of companies with gas tankers and dry vessels, with different noon reporting in the same company, Mr Lalechos said.

Mr Makris from Angelakos Hellas noted that different companies have different requirements for noon day report data, so there could not be one standard for all. For example, for cruise lines it is vital to report how much water they have onboard, but this is not so important in other shipping sectors.

Even within the tanker sector, companies want different data for VLCCs and MR tankers.  “Even charterers do not agree on the data that they want. Every charterer needs different data. So, there's not room for having common ground.”

Automated data collection

More automation in data collection and validation would improve performance data, said Mr Papandreou of Blue Autonomy.

Mr Lalechos’ company, NeptuneZero, has developed a system which generates a noon day report automatically on a vessel. It runs onboard, and does not require any internet connection.

Data gathered can include average power, running hours,  average kilowatts per engine, average RPM and turbocharging.

Crew like it because it makes their life easier, he said. But the systems can also detect if any figure has been altered before submission and alert the office.

Having systems which connect shipboard sensors to remote servers by satellite can be much more complex.

The sensor data figures are provided to the crew, who only need to add data about times of arrival, berthing, unberthing and anchorage. Then they give it a digital signature and submit it.

The software has APIs, so it can connect to other systems, including a data verifier or voyage optimisation company, or a company managing engine performance.

Software tools can verify the data using models, for example to detect if the tank should have 32 tonnes, but 36 tonnes has been reported. In such cases, a crewmember can be asked to check the reading.

These solutions can be “exceptionally low cost” compared to the first data collection systems which appeared in around 2010.

Mr Lalechos sees the pathway to automated data collection and management systems as similar to the pathway in business management from spreadsheets to software tools, with alerts and virtual assistants.

We could have systems which automatically send e-mails to the vessel to ask why something has happened, or which ask for evidence behind a piece of submitted data. They can send a reminder if it isn’t received in 7 days. It connects directly to the ERP and is logged as an “open issue”.

It means that vessel performance managers can spend their time troubleshooting and managing overall vessel condition, not just co-ordinating tasks. So we will still be reliant on people, but heavily empowered by tools.

Error reduction with monitoring

Having digital monitoring systems can motivate crewmembers to make sure data is correct, because they can see that digital systems might expose any errors, Mr Makris from Angelakos Hellas said.

Mr Makris told a story about when his company installed Metis, a system to gather and manage sensor data. On the first day the system was installed, a chief officer was required to take over a vessel because the captain was unable. He  decided to do a crash astern emergency stop, to check the systems work, two days prior to arrival in Brazil.

The vessel was 18 years old, and this emergency stop procedure damaged two out of six reversing mechanisms.  There were three spare systems onboard, which were installed without a problem.

But the chief engineer reported that the mechanisms had suddenly failed, without mentioning the emergency stop, something which seemed impossible.

Mr Makris looked through the Metis data and saw the crash astern cycle, so he knew the chief engineer was not telling the truth. “This was a turning point for our company,  everybody understood they have to provide accurate information.”

Working with performance data

The main purpose of gathering performance data is to find ways to use it to reduce emissions, not just to meet reporting requirements. Doing this needs something of a layered approach, said Mr Papandreou of Blue Autonomy​.

A starting point for analysis can be to follow the ISO 19030 standard method to measure changes in ship hull and propeller performance.  It’s transparent. “A 5 per cent speed loss on in-service performance [compared to the standard] is something everyone can understand,” he said.

Then, you need to analyse the data to understand the reasons for the loss of performance.

With the data, you might use predictive tools to forecast fuel use for a voyage, and tools to find the best route and speed. AI tools can give good predictability, he said. But you should use them together with domain expertise.

Then you might use prescriptive analytic tools to work out how to best optimise the vessel, choose the best options to save fuel or money.

It can be useful to compare similar vessels doing similar voyages. This can help motivate crew to improve performance. “Nobody wants to have the worst vessel in the fleet,” he said.

You can watch the panel discussion on YouTube at https://youtu.be/HDiHZEYVUjw