Atlantis Resources AK-1000: tidal turbine blade load testing and stab connector as-built

Atlantis Resources is a vertically integrated marine renewable power company. It provides highly efficient and reliable marine turbines, project origination, resource assessment, project installation and completion management, as well as operations and maintenance services to utilities and power...

Project at a glance

Client: Atlantis Resources
Sector: Marine renewable energy
Location: The UK’s first tidal blade test facility, and the pier at Peterhead
Date: Two phases, both delivered under Digital Surveys
Services: Deflection monitoring by 3D laser scanning, 3D deformation assessment, QA & Dimensional Control, as-built modelling, clash detection

The brief

Luminous (then trading as Digital Surveys) worked with Atlantis on a number of projects testing a prototype tidal turbine system they were developing, the AK-1000. Two of those projects are described here, and they sit at opposite ends of the same problem.

The first was the pressure testing of new blades at the UK’s first tidal blade test facility. We carried out various laser scans of the blades while under varying degrees of pressure and monitored the blades for deflections.

The second, and later, project was to create an as-built model of an existing turbine which had recently been brought up out of the water. “As-built” means the geometry as the object actually is, not as it was drawn. Atlantis required a very accurate model of the female stab connector, the interface through which the turbine mates with its subsea foundation.

The challenge

A tidal turbine blade is a large, doubly curved surface that flexes under load. Traditional deflection measurement attaches contact instruments at a handful of chosen points, and reports movement only at those points. If the blade deforms somewhere the engineer did not instrument, that deformation is not measured. On a prototype, where the deflected shape is exactly the unknown being investigated, that is a significant limitation.

The stab connector posed the opposite problem. Here the requirement was not movement but absolute fidelity. A mating interface has to be modelled tightly enough that the model can be trusted for clash detection, the process of checking new or proposed geometry against existing geometry to find physical conflicts before anything is built or installed. A model that is broadly right is of no use for that. Equipment recovered from the sea is also unlikely to match its drawings exactly, so the design model could not be assumed correct.

Our approach

For the blade 3d deformation testing, the method was a reference scan of the unloaded blade and rig, followed by successive scans at each load stage. Each load-stage scan was compared to the reference as a 3D deviation, giving a maximum deflection for that stage, with sections cut through each test. Because the comparison is surface to surface rather than point to point, the whole blade is measured at every stage, not just the places somebody thought to instrument.

For the stab connector, the turbine was located at Peterhead on the pier. We opted to use a Surphaser laser scanner. The Surphaser provides an unparalleled level of quality and low noise, which makes it ideal for engineering projects. Across the fleet these instruments are specified at range noise of 0.07mm at 10m and range uncertainty of 0.35mm at 5m, capturing up to 40 million points per scan, and we describe them in house as the most accurate tripod based scanner available.

3D Laser Scanning

We carried out 18 laser scans around the turbine. The data was then processed back in the office with each individual 3D laser scan aligned in Leica Cyclone using the cloud-to-cloud method. Cloud-to-cloud registration aligns separate scans into one dataset by matching the overlapping geometry between them, rather than relying on physical survey targets. Our standard practice is to accommodate at least 30% overlap between scans so that there is enough shared surface for the software to lock onto. Due to the incredible accuracy of the Surphaser data, cloud-to-cloud registration works very well: low noise means the surfaces being matched are sharply defined.

Once registered, the data was ready for modelling. Modelling was carried out in Inventor 2013 using the then new native point cloud support, which allows the point cloud to be referenced directly inside the modelling environment rather than converted first. Once completed, the model could be clashed against the existing design model to check for clashes or fabrication discrepancies.

What we found

The blade load testing produced a measured deflection at every stage of every test.

On Blade 1, Test 08, Scan 01 was the unloaded reference. Scan 02 returned a maximum deflection of 271mm, Scan 03 466mm, and Scan 04 469mm.

On Blade 1, Test 09, Scan 05 was the reference. Scan 06 returned a maximum deflection of 251mm, Scan 07 439mm, and Scan 08, taken unloaded, returned 2mm.

That final 2mm reading is the important one. With the load released, the blade had returned to within 2mm of its reference shape across its whole surface. The structure had absorbed nearly half a metre of maximum deflection and recovered. Elastic behaviour on that scale is exactly what a prototype blade programme needs to demonstrate, and here it is demonstrated over the entire surface rather than inferred from a few gauges.

It is worth stating plainly what the method achieved: half-metre deflections measured across a whole blade surface, at multiple load stages, without attaching a single contact instrument to the blade.

Deliverables

For the blade tests we issued per-scan deviation plots, each with a colour ramp showing the distribution of movement across the surface and a statistics block giving the numbers, plus section views for each test.

For the stab connector we issued the registered point cloud and the as-built Inventor model, ready to be clashed against the design model.

The outcome

Atlantis received quantified, whole-surface deflection results for its blade programme and a measured recovery figure confirming the blades returned to shape after load.

On the turbine, the as-built model gave the engineering team geometry they could trust for interface work, so clashes and fabrication discrepancies against the design model could be identified onshore rather than discovered during a subsea operation. That is the value of measuring what was built instead of trusting what was drawn.

Due to the incredible accuracy of the Surphaser data, cloud to cloud registration works very well.