Blog

Why you need trends when comparing ship motion data (and how you avoid blind spots)

Magpies are among the smartest birds and can even fool advanced scientific researchers studying them. There are many signs of their incredible intelligence. For example, they can remember dozens of distinct faces. They’ll use this memory to swoop attack anyone who has done them wrong. Even after decades have passed, too. So, researchers wanted to gain more insight into their behaviour. They planned to do this by tracking individual bird movements. To do this, they needed a tracker that would be impossible for a bird to take off. They spent months perfecting a harness design. And they seemed to have succeeded with a robust and simple solution. Yet once they started the experiment, things began to unravel almost immediately.

After attaching the devices to several birds, they eagerly started collecting data. But researchers were dumbfounded as, one by one, the harnesses on the magpies popped off. True to the design, no individual bird was able to take their harness off. But what actually happened was the birds cooperated with each other. They took turns pecking the harness off each other. The research hinged around a focus on an individual bird. But the birds’ ability to act together as a group left the researchers with a big blind spot.

Similarly, we can get blind spots like this in our work. This is especially the case when making comparisons. In the world of ship motion analysis, your research may hinge on comparison to an individual test case. However, it’s crucial you consider a group of cases together to get an accurate understanding. In this article, we’re going to cover why you need to focus on trends when comparing ship motion analysis to test tank results.

What do we mean by trends when making comparisons?

Collecting data from test tank facilities or field trials in sets is common. This data consists of multiple runs of recorded ship motion and acceleration data in as consistent conditions as possible. In a ship motion analysis, a run might consist of a recording of ship motion in waves of a certain height and period, with a similar relative heading to the hull, with the ship at the same consistent speed.

This kind of data is vital to ensure numerical models are truly accurate and meaningful. In this context, making comparisons merely means the difference between ship motion prediction calculations and the measured values. However, it’s the trends between multiple comparisons that provide a lot of crucial insight rather than the comparison of the individual runs.

Trends provide something that the individual runs do not: a normalization of uncertainty

Of course, the individual runs provide valuable insight on their own. However, trends are crucial because numerical models always have some underlying assumptions or simplifications. This means you should expect to get some differences from reality. These simplifications are necessary to get meaningful insight with a balance of practicality in the numerical computational cost. Even the most advanced seakeeping analysis programs have multiple uncertainties and limitations. So, what kind of uncertainties are we talking about?

One uncertainty comes from seakeeping programs ignoring the transient effects of wave-making and viscous drag resistance on the hull

Usually, hull resistance is characterized as a function of ship speed only, and this only comes up in numerical predictions of maneuvering or combined seakeeping and maneuvering analysis. However, it’s missing the transient effects, or in other words, how these physical effects ramp up or down as the hull accelerates in water. Yet transient effects are not the only area of uncertainty: how the flow moves relative to the hull and appendages is a concern.

Another area of uncertainty has to do with the angle between the relative flow to the hull and its appendages

The angle between the relative flow and the appendages is the angle of attack. When there is a high angle of attack on either appendages or the hull itself, viscous effects can get very complex. The characteristics of the way the flow moves can change. These changes in flow mean there can be dramatic changes to lift and drag on the hull and appendages. For example, hull appendages, like fins, may reach high angles of attack during large rolling motions at lower ship speeds. This can lead to stall and a complete loss of lift force on the appendages. In another case, too much vessel yaw during an extreme turning maneuver could lead to flow separation on the hull and a significant shift in the viscous forces on the hull. So far, the effects of viscous forces in time and space have been a substantial source of uncertainty. But some uncertainties come up in wave hydrodynamics, too.

Another uncertainty in seakeeping analysis is from significant changes in wetted hull area

Many numerical models of these effects assume a constant waterline area, allowing only for relatively small deviations from wave effects. This is why it is well worth recalculating ship hydrodynamics for different displacements. But the waterline area is dynamic in reality. In more extreme conditions, in larger and steeper waves, or when the ship is near a resonance condition in pitch or roll, the wetted area can change a lot. Yet these changes are also transient, especially since the nature of ocean waves and ship motion will always have some dynamic component. These are a few noteworthy examples, and though they may seem very different, they do have some common themes in terms of extreme motion – both in terms of velocity and position.

These uncertainties will be more critical during extreme motions

Extreme transients can mean moments of very rapid velocity or acceleration. These effects show up for some amount of time, but it might be variable depending on what’s happening during the run. Another theme is in extreme displacement – that is, during extreme limits of motion, or in sea states with extreme waves. This is important because it means they are less frequent in mild conditions and show up more often in moderate to extreme conditions. And this has everything to do with normalizing the uncertainty across comparisons to multiple runs.

There will always be a variation in the amount of these hydrodynamic uncertainties through each ship motion run

This is because of the way the ship moves in an irregular pattern through a sea state, with varying amounts of acceleration through time. Yet it simply isn’t possible to know in advance how many and how often these events will happen. It’s also common to see differences between degrees of freedom – roll being the most dynamic, and heave and pitch motion being less so, at least for monohull ships. And this is why a comparison of trends is so meaningful – it normalizes these uncertainties across more data. The trends then help you understand if you have the fundamentals correct. The blind spot comes in if you over-focus on any one individual run.

An individual run might have more errors than most

But there are a lot of variables in a ship motion simulation, and zeroing in on every hydrodynamic effect at each point on the hull as it marches through time relative to a wave can be inordinately time-consuming. You can spend a lot of time trying to understand the difference between a specific run of test tank or field trial data without getting much insight. In the meantime, more valuable insights can always come from looking at a trend of runs in the same conditions. It doesn’t mean you ignore any big differences completely, but at a certain point the trends are what’s useful in characterizing how accurate a model is. There will always be a spread in the differences between actual measured ship motion and numerical model results. Heave and pitch are often closer together, and roll has the most variability in differences because of the sensitivity to complex viscous effects.

Can CFD provide a more comprehensive answer to ship motion analysis?

Seakeeping analysis with Computational Fluid Dynamics (CFD) can capture transient hydrodynamic effects and complex viscous effects like drag and lift on the hull and appendages in high angles of attack. But CFD has its own set of numerical model uncertainties. Using CFD also opens up new questions on the sensitivity of the results to the mesh, the selection of the turbulence model, and other factors that require a lot of careful thought. Most importantly, the time and expense from the computational demand of CFD tools can make them impractical for use in seakeeping analysis, especially in the early stages of design when rapid feedback and iteration are required. CFD tools may answer some complex and transient flow effects, but there are other appropriate tools to use in some cases.

It’s time for an example

The best kind of validation is a comparison to full-scale ship motion measurements at sea. Let’s take a closer look at the comparison between the ProteusDS ShipMo3D toolset and measured motions of the HMCS Nipigon Destroyer in a moderate sea. The publicly available full-scale data incorporates a wide range of forward speeds and relative wave directions in moderate wave conditions for a comprehensive dataset.

You can see a listing of specific results comparing measured full scale ship RMS heave motion with seakeeping predictions. Individual cases range in difference from as low as

Generally, heave and pitch motions are within 10% of measured values, and roll is within 30%, which is typical for ship motion prediction software programs.

It’s summary time

We covered a few aspects of comparing numerical predictions and measured ship motion data, and it’s time to review. Measured ship motion data in the form of tank tests or field trials are invaluable – vital even – for developing and using numerical ship motion prediction software. Comparisons provide insight and confidence as to how accurate your ship motion predictions are. However, focusing too closely for too long on any individual run is a trap. There are always simplifications and assumptions in even the most sophisticated ship motion prediction software. This means there will always be some differences, and it can be immensely complicated and time-consuming to investigate individual runs. You still need to pay attention to any problems with the numerical model. However, spending too much time and energy on exploring individual cases can leave you with a blind spot if you don’t consider trends in comparisons. It’s about getting the bigger picture – a bigger picture which is straightforward to get because you don’t have to work with Magpies that mess with your testing equipment!

Next step

In the example, we used some of the data from the full-scale ship motion comparison between the HMCS Nipigon and the ProteusDS ShipMo3D toolset. Read more about the full-scale comparison in the ProteusDS ShipMo3D toolset validation report by Defence Research and Development Canada here.