Composite materials are making inroads in construction of equipment for the hydroelectric power industry. An investigation into material strength and other criteria reveals many more applications, particularly for small and micro units.
By Marc Whitehead and Roberto Albertani
Marc Whitehead is mechanical EIT (engineer in training) with HDR Engineering Inc. At the time he completed this research, Whitehead was a fellow with the Hydro Research Foundation. Roberto Albertani, PhD, is associate professor at Oregon State University and was Whitehead's advisor through the research discussed in this article.
This article has been evaluated and edited in accordance with reviews conducted by two or more professionals who have relevant expertise. These peer reviewers judge manuscripts for technical accuracy, usefulness, and overall importance within the hydroelectric industry.
The rise of new materials provides exciting opportunities for the hydroelectric industry. Wood -- used in the original waterwheels and penstocks -- was supplanted in part by steel components in the early 1800s. Steel retains its strength through high fatigue loading and resists cavitation erosion and corrosion. Its properties are well-understood and the processes for component manufacture are well-developed. For large units, steel will likely remain the material of choice.
However, given the rise of small (below 10 MW) to micro-sized (below 100 kW) turbines, composites can be used to save weight and reduce manufacturing cost and environmental impact. This is especially relevant given the continuing need for growth in electricity supply. The installed world hydro capacity, nearly 800,000 MW according to a 2009 study by Norwegian Renewable Energy Partners, is only 10% of the economically feasible and 6% of the technically feasible hydropower. The potential to bring more of the technically feasible hydro into the realm of economically feasible increases with the ability of composite components to provide economy of scale.
Modern fiber-reinforced composites have stability and fatigue strength that can rival steel components. The components considered in this study are constructed with a polymer matrix and synthetic or natural fiber reinforcement. They are chosen for their strength-to-weight ratio and economy of scale for complex shapes.
Current applications in hydro
Composites have found a place in demanding applications, including the body of airplanes, axle frames for rail cars and wind turbines. Composites also have been used in hydrokinetic turbines. For example, the 1.2-MW SeaGen tidal turbine being installed in Northern Ireland uses a combination of fiberglass and carbon fiber to produce stiff and lightweight blades.
Composites in hydroelectric turbines must tolerate harsher running conditions. Higher speeds can induce vibrations, and lightweight turbines may have lower resonant frequencies than heavier steel counterparts. Additionally, the epoxies used as a binder in composite turbines do not have nearly the cavitation resistance of steel turbines. Research has been completed on coatings to resist cavitation damage in steel turbines.1 Similarly, there is a study demonstrating that certain coatings can reduce cavitation erosion to similar rates to 316 stainless steel.2
Composite bearings are also finding applications in conventional hydro as thrust bearings for runners, wear plates and trunnion bearings on spillway gates.
Further development of composites in the hydroelectric market is proposed. While there are many issues to be considered in developing these components, those addressed here are technical feasibility in terms of strength and deflection and economic feasibility when compared with existing steel components.
The penstock, wicket gates/stay vanes, scroll case draft tube and runner for two turbines were selected for design and manufacture with composite materials. Steel designs were used as a starting point for design of the composite components. The specifications for the two turbines, a 2-MW Francis and a 250-kW propeller machine, were acquired from Canyon Hydro based on head and flow that kept the power in the range of small and micro hydro, H = 45.7 and 15 meters and Q = 4.5 and 1.91 cubic meters per second (cms). These designs provided the necessary dimensions for all the components, except the runner. Runner models were obtained from Voith Hydro and scaled to match the dimensions of the Canyon Hydro models.
|These runner models for the Francis turbine (left) and propeller unit (right) were obtained from Voith Hydro and scaled to match the dimensions of the entire unit models supplied by Canyon Hydro.|
Composite component design
Designing turbines from fiber-reinforced composites is complex because the materials are anisotropic, meaning they have high strength and stiffness in the fiber direction but are much weaker in the transverse direction, where only the matrix contributes to the mechanical properties. Plies of fiber are laid on top of each other in varying orientations to achieve the necessary mechanical properties, creating a composite laminate. The wicket gate, scroll case and draft tube were designed with a Quasi-isotropic layup. A layup refers to the combination of different plies (sheets of reinforcing fiber) at various orientations (directions). A quasi-isotropic layup orients the plies of carbon fiber such that the final layup closely resembles an isotropic material. The penstock is filament-wound and thus lacks longitudinal fibers, and the two runner models are made of an isotropic forged carbon fiber composite with randomly-oriented fibers in an epoxy matrix.
|Results of previous computational fluid dynamics analysis (left) and the finite element analysis performed in this study (right) show that stresses on the blades were within the tensile strength, but the factor of safety was less than 2 for significant portions of the blade.|
Penstock dimensions were selected based on the inlet diameter of the scroll case. As the penstock is a pipe, a quote for filament-wound fiberglass pipe was obtained for the closest standard size to the turbine inlet, 1.07 m. The size for both turbine cases was the same and the pressure rating was confirmed using manufacturer-supplied thicknesses and hoop stress.
Steel design was made for comparison, and thickness was limited by installation and handling.3
|Weights were calculated for the five turbine components for parts made from composite and steel materials. These numbers are based on 50% fiber weight.
The draft tubes were modeled as simple conical diffusers. The 250-kW unit had inlet diameter of 0.58 m, outlet diameter of 1.12 m and length of 2.29 m. The 2-MW model had inlet diameter of 0.861 m, outlet diameter of 1.64 m and length of 4.65 m. The pressure load at the runner exit was the design criteria. The load is a low vacuum due to head differential between the discharge ring and tailrace water level. The low structural requirements made the draft tube a perfect candidate for natural fiber reinforcement. Flax was chosen for its high strength, among natural fibers, and low cost. A factor of safety of 15 was applied to account for the uncertainty of using natural fibers and resin. The beam loading for a horizontally mounted full draft tube simply supported at both ends was also checked. Steel design followed the same thickness requirements as the penstock.
Figure 1 on page 56 shows models for both runner geometries. Finite element analysis (FEA) was performed and the Von Mises failure criteria were applied as the proposed composite material was a high strength, quasi-isotropic sheet molding compound (SMC). For this analysis, deformation was secondary to strength, and it was assumed further redesign would be required to accommodate a change in materials. Vibration analysis and more in-depth CFD was outside the scope of this research project. Loading on the turbine blade was calculated based on unit power. It was assumed the rated power was 90% water-to-wire, so the turbine power was increased by 10%. This power was then divided amongst the blades and divided by the speed of the turbine, as supplied by Canyon Hydro, to find the load applied to each blade. The load was then applied over the blade surface as a surface traction with a linearly decaying distribution to account for the higher loading at the leading edge. Literature on FEA analysis of turbines was checked to ensure the applied loads and resulting stress and strain were comparable.
Figure 2 on page 57 shows previous computational fluid dynamics (CFD) analysis alongside the FEA analysis performed in this study. Results showed that stresses on the blades were within the tensile strength, aside from overly stiff boundary conditions, but the factor of safety was less than 2 for significant portions of the blade, which is less than ideal for an unknown application of composite materials. It is outside the scope of this study to redesign the turbine blades, but it would only require minimal increases in thickness at the locations of highest stress to bring the factor of safety above 2. However, given the tight tolerances and low factors of safety applied to the manufacture of advanced composites in the aerospace industry, a lower factor of safety could be justified for this application as well. Deflections of the blade were higher than stainless steel given the lower stiffness of the carbon fiber random fiber.
The wicket gate was modeled as a simple symmetric hydrofoil shape. This provides an accurate base design to prove the feasibility. The height and depth were 0.24 m and 0.115 m for the 2-MW case and 0.20 m and 0.110 m for the 250-kW case. Loading was calculated by modeling the wicket gate profile as an airfoil with the structure modeled as an asymmetrically loaded beam. A factor of safety of 2 was applied to ensure there was adequate strength for operation outside of normal full load scenario, such as gate opening and closing and water hammer. The higher loads necessitated the use of higher-strength s-glass-reinforced epoxy in the design of the wicket gate. To make the process more streamlined, the wicket gates will be manufactured as a long airfoil section that can be cut into separate gates. The tops and bottoms can then be finished and pins attached to each piece separately into the body of the gate and bolted on through each side. The steel wicket gate used for comparison was modeled as a solid steel vane per industry standard design.
Deflections for the steel wicket gate were 0.00007 in and 0.0004 in for the composite wicket gate. It is assumed that the higher deflections will not cause sealing trouble during closing of the gates for the small size of the units considered in this study. More in-depth analysis for gate squeeze and shear pin failure was not included, as the primary focus was hydraulic strength and manufacturability of the composite components. It is assumed connection details would be resolved in future design and prototyping.
The scroll case was designed according to dimensions supplied by Canyon Hydro. The design load was the hydrostatic pressure of the 2-MW turbine model as cited by Hydraulic Design of Hydraulic Machinery.4 Given that the case has the same loading scenario as the penstock and is also a cylinder, the hoop stress equation was used on the limiting design diameter to determine the required fiberglass thickness. Steel thickness was the same as the penstock and draft tube.
Composite component manufacture
To manufacture the penstock economically and with consistent high strength, the best method is filament winding. A large mandrel is wrapped with tows of fiber that have been run through a resin bath. The tows are wrapped in hoop and helical patterns to create strength for internal pressure, longitudinal bending and handling. The results section below shows the cost and weight per foot for the two penstock sizes, based on a quote from local suppliers. The quote showed that the design thickness was driven by installation and handling requirements, rather than the relatively low pressure load, and for both it was 2.28 cm.
Two manufacturing methods were considered for the wicket gates and stay vanes; wet layup and vacuum infusion. Wet layup uses dry fabric, which is impregnated by pouring resin over the fabric and using rollers to push the resin into the fabric. This process is not as clean as vacuum infusion and does not always produce the most optimized structure in terms of fiber-to-resin ratio, but it takes less time than the vacuum infusion process. Vacuum infusion lays up dry fiber in the correct orientations, and the dry stack is then vacuum bagged and extra fittings are attached that lead to a resin supply, which is drawn into the part when the vacuum is applied. The vacuum helps maintain the amount of resin at an optimal level and reduces the release of volatile organics.
The scroll case will use a hand layup in two separate halves on a male mold to ensure a smooth inner surface. These two halves will then be bonded together with fiber added to the outside at the bonding point to ensure adequate strength. The pressure load in the scroll case does not require a high-strength advanced composite, so a wet layup of fiberglass fabric with an epoxy resin will be sufficient. The thickness of the scroll case was based on the same design parameter as the penstock. The 250-kW unit is an axial flow machine, so there is no scroll case.
A turbine runner combines a complex geometry with high load requirements. Recent work has demonstrated that high-strength structural components can be manufactured from a chopped prepreg SMC with excellent strength and stiffness.5 The suspension arm of the Lamborghini Gallardo was designed using multiple layers of a chopped prepreg SMC known as a forged composite, compression molded to produce the required thickness. The same method can be applied to the Francis and propeller runners. The Francis runner cannot be made as one unit, as the complexity of the blade overlap would prevent the part from being extracted from the mold. Thus, the runner blades, crown and band are manufactured separately and then bonded together and reinforced with bolts through the outside of the crown and band.
While the draft tube is most easily manufactured using filament winding, this process has not been commercialized using natural fibers. Thus, hand layup was chosen, as this is standard method of manufacture, despite the higher labor costs. Using a male mold similar to a mandrel, the layup can be completed with the mold horizontal and then turned vertical to cure, preventing sagging on one side. Figure 3 on page 58 shows the weights of the five turbine components for parts made from composite and steel materials. The weight of the composite parts will vary slightly depending on the amount of resin in the finished part. These numbers are based on 50% fiber weight.
The total weights for the steel and composite 2-MW turbine are 9,888 kg and 7,016 kg, respectively. The 250-kW steel and composite turbines are 3,734 kg and 1,927 kg, respectively. The totals assume 20 wicket gates for each turbine and a penstock length equal to the head of the turbine. It is likely that the penstock would be longer and require fittings, but this number gives a basic estimate of the weight of the unit and associated peripherals. The generator, bolts and gate actuating hardware are not included and are assumed to be similar between the composite and steel units. It is also worth noting that the runner redesign required to account for stress concentrations seen in the FEA would add weight to the composite units, but the amount is assumed to be minimal, on the order of 5 kg to strengthen points with stress concentrations.
With the given weights, the 2-MW composite turbine and its penstock could be lifted by the fast V-22 Osprey, whereas the steel machine would require a slower, less maneuverable Chinook twin rotor helicopter. Also, the 2-MW composite turbine and penstock could be towed by an F-250 4x4, whereas the steel unit would require a larger truck that would be difficult to maneuver on forest roads if the installation was remote.
A summary of the manufacturing and cost of each component is presented in Table 1 (on page 60) for the 2-MW turbine and Table 2 for the 250-kW turbine. The total cost for each component was calculated by dividing the mold among a 10-part run. While this is very small, it shows the cost for a small pilot run of composite turbines.
Process modeling for labor and tooling were taken from multiple sources. Northrop Corporation produced a widely used process model that covers hand layup of the wicket gates, draft tube and scroll case. The model converts measured labor time for activities in the manufacturing process into equations. The equations average time over multiple runs of the same process. This model was developed for wet hand layup and prepreg but was adapted for the other manufacturing methods in this study. To confirm the adaptations, case studies were used as a baseline. A study by wind turbine manufacturer Gurit was used to confirm labor cost for vacuum infusion of the wicket gates6 and a study from MIT for SMC forged composite method proposed for manufacture of the runner.7 Molds and material costs were estimated using these and other case studies, including process models developed in other academic research. Direct quotes from manufacturers were obtained to confirm calculated values and find prices for some items, such as the penstock, that were already commercially available. These commercially obtained numbers provided a comparison for the new components to ensure that estimated costs were within a reasonable range.
Weights of the solid components were based on the volume of material from the Solidworks model used in FEA. Hollow components were calculated based on the thickness of the composite skin and form core, as applicable. Steel components were based on a design thickness obtained from manufacturers (solid wicket gates) or design manual thickness (penstock and draft tube). Costs for materials were calculated using similar parameters and combined with the labor and tooling costs and confirmed by the process models noted above.
It is feasible to construct turbines from composite materials, and a weight reduction of 50% to 70% was seen compared to conventional steel components. The reduced weight can allow composite turbines to be installed in remote locations. In addition, assembly of these composite structures does not require welding equipment. The components also require fewer parts to be bolted together, as each piece can be made in one or two sections. At the small production runs modeled in this study, the cost of the molds and other tooling dominate the component cost.
The small runs indicated here show what it would cost to begin further research into these materials. This research can address cavitation erosion and UV protection of the components after installation. It may be possible to use elastomer or ceramic coatings to reduce cavitation or ensure that the turbine runs in the flow and head regimes that prevent cavitation from occurring. It will be important to test and resolve these and others issues to ensure the units can achieve similar reliability to steel turbines, especially if they are to be installed in areas where maintenance will be infrequent.
Even at these small runs, some composite components can be cost-effective due to the decreased labor required for manufacture. For example, a scroll case for the 2-MW Francis unit would cost $80,000 to be welded from steel compared to $25,000 for composite manufacture. However, assuming successful design of turbine runners, the cost for molding the composite runners is more than equivalent steel components. The 2-MW runner would cost about $23,000 to manufacture from steel, compared to $27,000 from composite. Costs may vary by machine. And the cost for composite components would drop considerably at higher production runs if molds could be reused.
Researchers have already investigated the construction of turbine runners from composite materials.8 However, this study did not address cavitation erosion and the feasibility of construction. The next step for composite turbines is to design and build a scale model that will allow proof of feasibility and economy of manufacture. This unit can then be tested to determine efficiency and applicability, as well as methods for preventing excess cavitation erosion.
The author thanks Dr. Roberto Albertani for his guidance as an advisor through this research, Oregon State Universityfor providing the facilities and education that aided the work, and the Hydro Research Foundation for providing the funding that allowed him to take on and complete this project.
1Sollars, R., and A.D. Beitelman, "Cavitation-Resistant Coatings for Hydropower Turbines," U.S. Army Corps of Engineers, 2011.
2Light, Kendrick, "Development of a Cavitation Erosion Resistant Advanced Material System," Masters Thesis, University of Maine, 2005.
3Steel Penstocks, American Society of Civil Engineers, Reston, Va., 1993.
4Hydraulic Design of Hydraulic Machinery, Gower Technical Press, Avebury, United Kingdom, 1997.
5Feraboli, Paolo, et al., " ‘Forged Composite' Technology for the Suspension Arms of the Sesto Elemento," Lamborghini (unpublished), 2010.
6"Comparative Cost Study of Infusion versus Hand Lay-up," Wind Energy Handbook, Gurit Composites, 2013.
7Busch, J.V., "Primary Fabrication Methods and Costs in Polymer Processing for Automotive Applications," Massachusetts Institute of Technology, 1983.
8Wang, J.F., P. Janusz and N. Muller, "A Novel Design of Composite Water Turbine Using CFD," Journal of Hydrodynamics, Volume 24, 2012, pages 11-16.
More HR Archives Issue Articles