This special post shows the development of prefabricated UHPC floating farms (we will call them “rafts” in advance) to solve the problems of durability of the traditional structures, made with eucalyptus beams submitted to fatigue and under an XS3 exposure class. The research and industrial scale up carried out is explained through the nine Technological Readiness Levels (TRLs) proposed in the H2020 program of the European Commission (EC) .
UHPC was proposed because it allows to design highly prestressed hollowed beams that can minimize the weight of the structure and assure its adaptability to the swell while remaining in uncracked state. The UHPC beams were designed to have the same bending strength and stiffness than the traditional beams tested experimentally. Laboratory tests and full-scale prototypes of up to 540 m2 tested in real environment proved that the prefabricated UHPC beams, their connections and the assembling procedure are suitable, while a remote monitoring system is still proving the better performance of UHPC beams compared to ordinary concrete reference beams. Thanks to the savings of the UHPC rafts in maintenance and reparation costs compared to traditional solutions there are currently 18.669 m2 of UHPC rafts distributed in the Atlantic, Mediterranean, Baltic and North Sea basins.
1. TRL Introduction: Why a new solution was needed?
In the period 1990-2016, the volume of mussels produced in the EU has dropped a 20% , in contrast with the trend worldwide. One of the reasons is a progressive decrease of competitiveness due to the low degree of innovation in the sector. A major example of this is Spain, the largest producer in the EU. The outstanding water conditions allow to use rafts, an intensive farming system that concentrates 500 ropes of 12-m length each in only 540 m2 to harvest yearly approximately 80 t of mussels. Made of eucalyptus wood since the 60’, traditional rafts have three main negative consequences:
1) Economic: They have reduced and uncertain lifespan, between 10 and 20 years, requiring periodic maintenance and frequent reparations. The partial or complete collapse of a raft implies a production shutdown and the loss of harvest.
2) Industrial: Traditional rafts are built manually and inefficiently through a high-risk job in the inter-tidal zone using hammers and nails. The replacement of damaged elements is slow, complex, risky, and environmentally uncontrolled.
3) Social and environmental: They require the deforestation of high ecological value trees and the yearly use of paints that may drop on the estuaries
Figure 1. General and detail views of wooden rafts under service in Galicia
In the period 2000-2015, few novel designs of polyethylene rafts were proposed by different innovators to try to solve these problems and with poor results, showing excessive deformations (see figure below), high slipping surface for the farmer and high prices.
Considering that the UHPC performance allows to design durable and light structures, the four founders of the company Research and Development Concretes (RDC) proposed a raft which grid was made with prestressed UHPC beams to launch to the market a resilient and durable floating platform that:
1) Minimizes the operational expenses of the mussel farmers;
2) Helps the sector to progress towards the Sustainable Development Goals (SDG) of the United Nations , using a tool that is respectful with the life below water (SDG14), promotes the harvest of sustainable, affordable and not-fed protein (SDG2, 12) and supports the recovery of an EU productive sector (SDG8).
Figure 2. Polyethylene raft the day of its commissioning (left) and after three years under service (right)
2. UHPC rafts: From idea to market
The development of the UHPC rafts between 2015 and 2019 is explained through the Technological Readiness Levels (TRLs, Table 1) proposed by the EC , which measure the maturity of certain technology from 1 (basic principles observed) to 9 (actual system proven in real environment).
Table 1. Particularities of each TRL. Adapted from 
2.1. Concept (TRL1-TRL3)
Since its development in the 80’ , UHPC has proven to be competitive when the structure or element is under one or several of the following three scenarios:
1) Under environments or conditions where high durability provides a competitive advantage, as structures under XS exposure classes (EN206-1, ), fatigue, abrasion, thermal cycles, or impact loads. In these cases, using UHPC generally minimizes maintenance costs and increases the lifespan. Examples of this are the beams used in the cooling towers of Cattenom and Civaux power plants , or the strengthening of the Caderousse and Beaucaire dams and the channel bridges over the Maurinenne motorway in the Alps .
2) In situations where lightness provides a competitive advantage, as in floating structures, cantilevers, or elements with significant transport and handling costs. In these cases, using UHPC generally provides the most cost-efficient solution from day one, or is the only solution viable. Examples of these applications are the perimetral UHPC pontoons recently installed in the largest floating photovoltaic farm installed in a EU dam (5 MW, Alqueva, Portugal. Client: EDP), or the use of thin UHPC panels to stiffen modular buildings.
3) In singular structures, where UHPC provides slenderer solutions that the client may perceive as exclusive, paying for this uniqueness. Two known examples are the roofing of Montpellier-South of France TGV Station  or the construction of the MUCEM in Marseille .
In many applications UHPC is the best alternative because both durability and lightness provide advantages, as in footbridges [10, 11], in the road overlays mainly used in Switzerland and the US [12, 13], or in singular in-situ applications, as the reinforcement of lighthouses carrying the fresh UHPC by helicopter . Figure 3 shows a qualitative ternary diagram with the most common UHPC applications and to what extent durability, lightness, or slenderness are making of it the most competitive alternative.
A preliminary analysis carried out by RDC (TRL1, observation of basic principles, 2015) of what benefits could UHPC provide replacing the wood in the rafts showed that the three features mentioned previously provide significant advantages, as they require: 1) High durability due to the XS3 exposure class, abrasion, toughness to resist the potential impact of the boat, etc; 2) Lightness because the grid of beams floats thanks to six steel tanks protected with Fibreglass Reinforced Polyester (FRP). The lighter the grid, the lower are the cost of the tanks, the hydrodynamic forces, and the area unusable to harvest with the ropes. 3) Slender beams to reduce the stiffness in the axes of the plane of the grid, avoiding the sudden pull of the ropes that swell may provoke, and the loss of mussels associated to them. This showed conceptually that a competitive design with UHPC may exist, and the team started to formulate a predesign (TRL2).
Ternary diagram (E. Camacho, 2020) of competitive uses of UHPC.
A conservative predesign was done replacing the wooden beams by rectangular massive UHPC beams (section 30×30 cm) to estimate the order of magnitude of the cost and the weight, assuming as a predesign rule 1 €/kg of UHPC (approx. cost in year 2015). The grid of a traditional raft (Figure 3) was replicated with the same number and length of elements to facilitate the transition from wood to UHPC for the future clients.
A traditional wooden raft is a grid of 20 x 27 m (540 m2) formed by the elements indicated in Table 2, where the joists rest orthogonally on the secondary beams, and they lay on the primary beams perpendicularly. The typical weight of the grid of a wooden raft with floaters is 54 t. The provided that a UHPC grid would weight 85 t and have an estimated cost of 85.000 €, values 50% higher than for the wooden grid. Being the comparison in the same order of magnitude, RDC proceeded to develop the concept/design (TRL3, 2016).
The conditions for this design were: 1) Minimize the volume of UHPC of the beams to reduce the weight, cost, and environmental impact, 2) As the element length (27 m) requires a prestressing, the UHPC should provide at 18 h the required compressive strength to release the strands, 3) The beams should resist their manipulation during the assembling from a single lifting point in the center of the span, which is the worst-case handling scenario. 4) The design of the connections between beams to work as a grid should avoid the use of in-situ concrete to facilitate industrialization and enable the decommissioning and reuse of the elements. 5) The sectional design of the beams should be rectangular to resist the significant bending moments in both axes, and to make them similar to current wooden logs to reduce the fear of change of the potential clients. The section was massive in the D-regions and with hollowed core in the B regions (without connections). 6) As the wooden beams of the traditional rafts do not experience bending failures under service and their deformation with the waves is satisfactory for the clients, the UHPC beams were designed to achieve the same maximum bending strength and stiffness than wooden beams.
To replicate the response of the wooden beams, their capacity was obtained doing full-scale laboratory tests to primary and secondary beams extracted from a traditional raft decommissioned. Four-point bending test with a ratio span/depth=10 were done at the Concrete Science and Technology Institute of the Universitat Politècnica de València . Beams were submitted to a process of water saturation during the 4 days prior to the test to simulate the real conditions. The spans ranged between 3 m and 3.4 m and the sections of the eucalyptus beams between 300 x 300 mm (the two secondary beams tested) and 340 x 340 mm (the two primary beams tested), representing the variability of the real elements. The load was applied at a speed of 0.5 kN/s, measuring the and the load-deflection curve. The standard deviation of the equivalent bending moment of the four wooden logs tested did not exceed 20%.
At the same time, three hollowed-core rectangular UHPC beams were designed with 250 mm of width having different depths (200, 230 and 250 mm) and 12 prestressing strands. The minimum cover was limited by Table 4.203 of the norm NF P 18-710 , which determines that for prestressed UHPC elements, exposure class XS3 and structural class S3 (agricultural structure, following the EC2), the cover for a lifespan of 50 years should be 25 mm. Thus, the wall thickness was fixed in 60 mm, with a cover of 25 mm and a 0.6’ strand at mid distance. The theoretical load-deflection curve of these three beams was obtained considering the UHPC properties shown at Table 3, and they are compared at Figure 4 with the experimental results obtained from a secondary wooden beam. The results suggest that the UHPC beam with section 250 x 230 mm (width x depth) was the most similar to a secondary wooden beam in terms of maximum bending strength and deflection capacity, while a UHPC beam with section 350 x 230 mm provided equivalent response than a primary wooden beam. Thus, these sections were selected as the optimum ones, reducing the total thickness of the grid from 65 cm (wooden raft) to 45 cm (UHPC raft).
Figure 4. Left: Four-point bending test of a eucalyptus wooden beam. Right: Bending response of a wooden element tested and theoretical response of prestressed UHPC beams with different depths
Regarding the joists, three designs were proposed (Figure 5): 1) Using the same wooden joists that are used in the traditional rafts, hammering them on polyethylene prisms embedded in the precast UHPC beams (named Formex®Mixta). This is the most economic and lighter solution. 2) Using reinforced UHPC joists with surface texture to avoid slipping (named Formex®Delta). Their connection to the secondary beams is with bolts through polyethylene elements embedded in the joists. This solution without wood is oriented for fresher waters, where wood decomposes faster. 3) The UHPC joists are integrated with each couple of secondary beams to create precast frames. This model, named Formex®Plus, faces the farming applications that require a flat area.
Figure 5. The three different types of joists designed for the UHPC rafts (wooden, UHPC joists, and UHPC frames)
Options 2 and 3 avoid any wooden element and the need of applying any maintenance coating. In them, the dimensions of the UHPC joists were defined for providing the same capacity as the wooden joists (100×80 mm), which were submitted to flexural tests with 2-m of span (Figure 6). A UHPC joist with similar section and four bars of ɸ=16 mm has the same bending capacity, but they were increased to 120×80 mm to facilitate the pouring of the concrete and allow a safer walking on them during the farming operations.
Figure 6. Four-point bending test of a wooden joist (left) and comparison between them with a UHPC joist (right)
The test of the wooden elements shows a clear size-effect, having ≈50 MPa in the 300×300 mm beams and ≈95 MPa in the 100×80 mm joists. This proves that the higher the depth, the more efficient is to replace wood by UHPC. Table 2 compares the weights and dimensions of all the elements designed with UHPC with the wooden ones. The weight of a UHPC matrix (primary + secondary + edges) is 40% higher than a wooden one (54 t vs. 39 t), while the lightweight of a complete Formex®Mixta is 43% higher than for a wooden raft (92 t vs. 64 t).
Table 2. Comparison between the elements of a wooden and a UHPC raft
The connection between floaters and primary beams in the traditional solution is done setting the beams on the floaters and embracing galvanized steel sheets against their steel bolts. The same system is proposed for UHPC beams due to its high toughness. The connection between secondary and primary UHPC beams cannot be done with drilling and nailing as done in the traditional rafts, so the alternative is precasting the beams with notched polyethylene cuboids integrated vertically in precise positions. During the assembling of the raft the beams are positioned, drilling their cuboids along their depth, and crossing them with a galvanized bolt (Figure 7). This procedure has a tolerance of 40 mm in each direction.
Figure 7. From left to right: Detail of the connections in the UHPC beams. Drilling and screwing during the assembling.
After verifying with different stakeholders that this design with UHPC could fulfil the market needs, a Utility Model protecting in Spain floating structures to harvest molluscs made of fiber-reinforced concretes was granted in March 2016 (Publication no.: ES 1147609 U).
2.2. Prototyping and demonstrations (TRL4-TRL7)
TRL4 is the testing and validation of the components (beams and connections) at laboratory scale, determining if they can work together as a system. First, a full-scale prototype of the critical region of the connection was made to validate that the dimensions were appropriate to pour the fresh UHPC between the reinforcement and the formworks, and to allow the connection between the precast elements (Figure 8, left). A full-scale reinforced UHPC frame integrating beams and joists was done (Figure 8, centre) to verify the capacity to pour in one step this element.
Full-scale prototypes of two UHPC primary beams and one secondary beam were done to:
1) Validate the capacity to produce the 20-m and 27-m precast prestressed beams at industrial scale. A 1 m3 planetary mixer was used mixing 0.5 m3 per batch and pouring the fresh UHPC on two molds with the prestressed strands. The filling was adequate, with good flowability between the strands, the formwork, and the hollowed core EPS, showing the capacity to produce the beams avoiding cold joints between the batches. The time required to pour each batch was 20 min, filling the quality control prisms in the last part. The fresh and hardened properties of this first material characterization at industrial scale are shown at Table 3. The mechanical and durability values measured fulfill the requirements that the NF 18-470 indicate that UHPC should satisfy. Considering the XS3 (splash and spray of sea water) and XA3 (faeces of the seagulls have pH 4) exposure class of the application, the verification of the durability parameters was particularly relevant.
Figure 9. From left to right: Porosity test, migration test and measurement of the slump flow
Table 3. Characterization of the UHPC used to produce the beams
4-point bending tests (un-notched) were carried out to nine 100x100x500 mm prisms (L/h=4.5) prepared with the UHPC used to cast the beams. An inverse analysis specifically developed by López  was carried out to determine the residual strength ft,u and verify its capacity to comply the tensile requirements.
2) Prove that the strands can be released after 18 h. For this is required that fcm, 18 h>72 MPa, so that this value per 0.6 is higher than the compressive stress supported by the UHPC after the release, which is 43 MPa for the secondary beam. In all cases the values exceeded 50 MPa after 18 h, proving that the release of strands does not limit the production.
3) Verify experimentally that in a beam accidentally manipulated from a single central point a macrocrack does not appear. Under this test (Figure 10), the UHPC reached the strain-hardening state and microcracks had an average crack size of 20 μm. Due to the high level of prestressing, these microcracks could not be found with microscope after leaving the beam.
Figure 10. Left: Test of carrying the beam from a single central point. Center: Multi-microcracking next to the central section. Right: Crack width measured with a digital microscope (21 μm), magnification of 55x
4) Apply a cyclic-load test of the connection, loading a cantilever of the secondary beam connected to the primary. This full-scale validation of the connection (Figure 11) simulates the cyclic effect of the swell on the raft. The connection between the primary beams and the supports was fixed, reproducing how they are connected to the steel floaters in the rafts. The test consisted in increasing the loads alternatively in cantilevers 1 and 2 until reaching the maximum load expected under service (65% of the maximum bending strength of the beam). Previously, cantilever 1 was pre-loaded to 100% of this value to determine if the response of the beam is different if it has reached before the maximum expected service load.
Figure 11. Scheme (left) and test (right) of the connection between the primary and the secondary beams
The same test was done to wooden beams extracted from a traditional raft. The results (Figure 12) for the UHPC system show that Cantilever 1 (preloaded) has after each cycle a neglectable residual deformation, while this value is reduced for Cantilever 2. In the case of the wooden beams this value is significant for both cantilevers, proving that the stress concentration of the steel bolt on the drilled beams generates a progressive degradation.
Figure 12. Bending response of the secondary UHPC (left) and wooden (right) beams submitted to the cyclic loads
5) Compare experimentally the toughness and maximum bending strength of the UHPC and the wooden primary and secondary beams carrying out four-point bending tests (un-notched). The configuration of the tests is shown at Figure 13.
Figure 13. Up: Scheme of the bending test carried out to the beams. Down: Wooden beam submitted to the test
The bending response of the beams is shown at Figure 14. UHPC beams experienced a compressive failure in the head of the beam (Figure 15) when the strands were submitted to 90% of their tensile capacity, proving the efficiency of the design.
Figure 14. Bending test carried out for the UHPC and wooden primary beams (left) and secondary beams (right)
Figure 15. Left: Central span of the UHPC beam before starting the test. Center and right: Region after the collapse
6) After the destructive bending tests, the UHPC beams were cut next to the mid-span macrocrack to count the number of steel fibers in the section and determine indirectly the fiber orientation. This is done assuming that in a concrete with 2% of fibers in volume, approximately a 2% of the area of any section analyzed is steel. As the area of the section of a perfectly aligned fiber unit is known (Πr2), the number of fibers n1that should have the region studied if all fibers are perpendicular to it (η=1) is known. Using an image analysis software, the number of fibers in the region studied can be determined, and the ratio provides the fiber orientation factor η of the element. In the case of the beams, the value was measured in four sections and the average η was of 0.56 with a standard deviation of 5%. This value corroborated that with the pouring process used the fibers were aligned almost randomly (η=0.5), being this 3D orientation favorable to avoid cracks parallel to the longitudinal direction that could appear due to the absence of transverse rebar reinforcement.
The validation of the full-scale components in laboratory completed TRL4. TRL5 is the validation of the (sub)system testing it in an environment that simulates or is similar to the operational environment (so-called relevant environment). Considering the need of simulating the moorings, the meteo-oceanic conditions and effect of the ropes, the only representative test was directly floating a full-scale beta prototype in relevant marine waters to carry out a long-term demonstration of the technology (TRL6). This was done through a 13.5×27 m prototype purchased by the scientific and technological center AZTI and installed in Mutriku (Vasc Country, Spain) in October 2016, becoming the first world application of UHPC in aquaculture. AZTI is still providing periodic feedback of the state of the structure, which is yet (August 2022) harvesting molluscs and did not require maintenance or reparation. The structure was a Formex®Plus, formed by five lines of secondary UHPC frames set perpendicularly on three UHPC primary frames, each one connected to a couple of steel floaters. TRL6 allowed to validate the transport in a special truck, commissioning, launching, and mooring. Submitted to a significant wave (hs) ≈1.5 m in the externa part of Mutriku port and with water depth (d) of 12 m, the appreciations made were: 1) The structure showed good stability under service, working the farmers comfortably on it. 2) The texturized surface avoided the danger of slipping. 3) Ropes tied to the joists suffered a progressive break due to the hardness of the UHPC, showing the necessity to smooth their vertex for the next design. 4) After few months, the rust of the fibers in contact with the surface appeared. This has only aesthetic implications, which are not relevant in this sector. Biofouling was not found.
Figure 16. From left to right: Installation of the secondary frames; launching of the raft, and raft under service
As Galicia (Spain) concentrates 90% of the EU rafts, its waters (1.5<Hs<4 m; 12<d<38 m) were considered the operational environment to carry out the system demonstration (TRL7, 2017) with two UHPC rafts (Formex®Mixta and Formex®Plus, Figure 17) with similar plan dimensions than the traditional rafts (20 x 27 m, 540 m2). These prototypes were cofounded by the H2020 project SELMUS nº738777 and improved with the lessons learned from TRL6. Each one was assembled by three operators in two days in a ramp in the seaside to use the rising tide for the floating, avoiding the costs of renting a 500 t-crane.
Figure 17. Formex®Mixta in wavy (left) and calmed (center) waters, and Formex®Plus (right). Ria de Arousa, Galicia
Each structure integrated a monitoring system to also demonstrate the durability in operational conditions. It provided the potential and intensity of corrosion (icorr) of the steel strands of the UHPC beams, comparing it with a reinforced H25 reference with the same cover (Figure 18). Up to date the corrosion of the strands has not started, while in the reference corrosion is under propagation. More info can be found at .
Figure 18. Average intensity of corrosion of the steel strands UHPC beams) and the reference in the Formex®Mixta
These rafts are still under operation and with perfect conditions. Their testers (current owners) were the first prescribers of the resiliency and minimum maintenance of the UHPC rafts, showing that prototyping can be an efficient tool to demonstrate the UHPC potential and build the demand.
2.3. Industrial production and commercialization (TRL8-9)
TRL8 was reached validating and certifying the next two UHPC rafts produced under the conditions of the mass production, which started in 2018 in the UHPC precast factory PREFFOR (Vilamarxant, Spain). This included the obtention of the CE marking for structural lineal precast concrete elements, and the elaboration of the production procedures (prestressing, mixing, pouring, loading in the truck, etc) and an industrial quality control plan for the UHPC following the EN 206-1 . The last showed through a normality test the Normal Distribution of the UHPC production, with a fck=138 MPa for year 2019 (L=100 mm, 28 days, and air curing, Figure 20). A massive campaign of flexural tests allowed to characterize the tensile properties of the UHPC (160 kg/m3 of steel fibers) for different η orientation factors.
Figure 20. Left: Quality control of the of compressive strength of the industrial production. Right: Residual strength (ft,u) of the industrial production vs. fiber orientation factor in the section that suffered the bending failure
The steady state of the production and the success installing and using these first UHPC rafts under the different operating environments completed the process towards the full commercial deployment (TRL9). To support it, a comparative cradle to grave Life Cycle Assessment (social, economic, and environmental) was carried out between the wooden and UHPC solutions, showing significant benefits of the second option mainly due to the higher lifespan and the minimum maintenance needs . The number and type of UHPC floating farms installed until August 2022 are shown at Table 4.
Table 4. Number and total area of the UHPC rafts installed in the EU (39 rafts, 18.669 m2 in total)
Despite that the cost of an installed UHPC raft is approximately 50% higher than for a wooden raft, it is becoming the prevailing trend because it has proven the following benefits: 1) Unlike wood, maintenance or replacement of beams have not been required yet and degradation is not observed. The lifespan expected is of 50 years. 2) All UHPC beams have the same properties, reducing the rolling under a storm and avoiding the overloading that damages the stiffer beams in the wooden rafts. 3) Surface texture, flatness, and easiness to walk on the UHPC raft makes of it the best option in terms of health and safety.
Summarizing, the complete development of UHPC rafts through the 9 TRLs required 4 years, being the full-scale prototypes under operational environment (TRL7) particularly relevant to prove the durability of the system and foster the demand. Minimum maintenance costs, sailing stability, resiliency, and flatness are other of the benefits that are favouring the market penetration of this solution at the expense of wooden rafts. This first world application of UHPC in aquaculture proves the wide possibilities that this material offers for marine and offshore structures, as both durability and lightness provide significant competitive advantages.