Abstract
In order to reach 40% efficiency, sub-MW turbines must operate in a recuperated gas Brayton cycle at a turbine inlet temperature (TIT) above 1300 °C. Current sub-MW turbines have material-related operating temperature limits. Still to this day, there is no cost-effective rotor design which operates at such high temperatures. This paper introduces a novel, sliding-blade, inside-out ceramic turbine (ICT) wheel configuration, which could enable high-efficiency sub-MW recuperated engines to be achieved with cheap monolithic ceramic blades. The inside-out configuration uses a rotating structural hoop, or shroud, to convert centrifugal forces into compressive blade loading. The sliding-blade architecture uses a hub with angled planes on which ceramic blades slide up and down, allowing to match the radial expansion of the structural shroud. This configuration generates low stress values in both ceramic and metallic components and can achieve high tip speeds. A prototype is designed and its reliability is calculated using cares software. The result is a design which has a single blade probability of failure (Pf) of 0.1% for 1000 h of steady operation. Analyses also demonstrate that reliability is greatly dependent on friction at ceramic-to-metal interfaces. Low friction could lead to acceptable reliability levels for engine applications. The prototype was successfully tested in a laboratory turbine environment at a tip speed of 350 m/s and a TIT of 1100 °C without any damage. These achievements demonstrate the robustness of the sliding-blade ICT configuration. Further research and development will focus on increasing tip speed and TIT to higher values.
Introduction
Microturbines—sub-MW gas turbines—are a competitive solution to other power units, especially in mobile applications, because of their high power density, reliability, and low maintenance requirements. Gas turbines are also fuel flexible which makes them particularly suitable for future applications requiring lower emissions levels. On the downside, microturbines in a simple Brayton cycle configuration suffer important losses when increasing the operating pressure ratio, counteracting other possible advantages [1].
When operated in a recuperated Brayton cycle, microturbines have the potential to reach 40% thermal efficiency while keeping their other advantages. To do so, microturbines must operate at a low pressure ratio (between 5:1 and 10:1) and at a turbine inlet temperature (TIT) around 1300 °C and above [2,3].
In the last decades, the TIT of large gas turbines (>1 MW) has dramatically increased through innovations in cooling technologies, which also drove turbine component prices to high levels. These technologies are hard to implement cost-effectively in microturbines.
In an attempt to solve both efficiency and cost challenges of microturbines, tremendous development efforts have been made to develop monolithic ceramic gas turbines. Since the 1970s, monolithic ceramic properties, cost as well as manufacturing processes have significantly improved thanks to many substantial development programs. With these improvements in material quality and reliability, some programs successfully tested engines above 40% thermal efficiency, but never reached the required reliability and foreign/domestic object damage resistance to become products [4–6].
The failure of these massive research programs indicates the need for a drastic design change for monolithic ceramic turbine to succeed. An interesting and largely unexplored solution to solving the issues of monolithic ceramic turbine is the use of an inside-out configuration.
Monolithic ceramics have high compressive strength, but low tensile properties. It is thus desirable to design monolithic ceramic parts that undergo compressive loading rather than tensile. The idea behind the ICT is to exploit the high compressive strength of ceramics by clamping ceramic blades between a flexible hub and a rotating structural shroud. In operation, the centrifugal force presses the blade against the structural shroud leading to a ceramic blade loaded mostly in compression. This configuration leads to a much lower failure probability of the ceramic blade and a higher blade resistance to foreign object damage/domestic object damage. The only current material candidate for the structural shroud, capable of withstanding the high rotational speed as well as the centrifugal load from the blades, is carbon fiber reinforced polymer. Since carbon fiber reinforced polymers cannot resist high operating temperatures, cooling is required to insulate the structural shroud from the turbine's hot main flow.
The inside-out architecture was first introduced by Kochendörfer et al. [7] in the 1970s with an all-ceramic rotor surrounded by a carbon fiber rotating structural shroud. The use of a ceramic hub and a ceramic cooling ring led to poor reliability, but still demonstrated the interest in this novel architecture.
More recently, inspired by the R4E engine [8], an ICT rotor comprising a superalloy flexible hub and cooling ring was developed and tested [9]. The prototype was tested at high temperature with no significant failure. However, tip speed limitations and crack propagations at the bottom of the blade were noted by the authors as challenges to address. Two major issues emerged from the design of a radially flexible hub for an ICT: (i) high tensile stresses in the hub and (ii) a high pressure interface at the bottom of the ceramic blade which results in local tensile stresses in the blade.
To address these issues, a sliding-blade turbine concept was proposed by Landry et al. [10]. The first step of this architecture exploration was the validation of the rotordynamics of the architecture, which was carried out successfully. It was concluded that rotordynamics stability in this configuration requires contact between the blade and the sliding plane in all conditions. However, the concept was not designed for nor tested under high-temperature turbine operating conditions.
This paper presents (1) the design of a sliding-blade ICT wheel for high temperature operation, (2) the experimental validation at a TIT of 1100 °C and a tip speed of 350 m/s, and (3) the ceramic blades life expectancy analysis. Experimental results and life predictions show that, with incremental improvements, the sliding-blade inside-out ceramic turbine concept is a viable architecture to significantly increase the TIT over state-of-the-art microturbines, providing gains in both efficiencies and power density.
Sliding-Blade Turbine Design
The sliding-blade architecture eliminates the source of both hub and ceramic blade tensile stresses that were problematic in previous ICT designs by using a rigid hub with inclined planes, which serve as tracks for the blades to slide up and down. Contact between the hub and blade sliding planes is achieved with the axial force applied by springs to the blades. The friction between the blade and the structural shroud, combined with the force applied by the springs, maintains the integrity of the turbine in all conditions. The prototype design is showed in Fig. 1.
Blade motion is dictated by expansion of the structural shroud, due to centrifugal loading. As rpm increases, the shroud radius increases, and the blades go up their respective sliding planes, as shown in Fig. 2. The sliding-blade architecture uses an axial spring comprised of a star-shaped disk with one spoke per blade. This ensures that each blade receives independent force and maintains contact with the hub when moving along its sliding plane, which is key to maintaining the structural integrity of the turbine wheel.
Displacement of the sliding plane architecture at rest (left) and of the sliding angle at full rotational speed with shroud fully loaded (right)
Displacement of the sliding plane architecture at rest (left) and of the sliding angle at full rotational speed with shroud fully loaded (right)
Prototype
In order to test the proposed design, a functional prototype was built, as shown in Fig. 3. The prototype is made of:
A structural shroud made from filament wound carbon fiber reinforced polyimide with 1800 MPa ultimate hoop strength and a 335 °C glass transition temperature made by Proof Research Advanced Composites Division (Dayton, OH).
Axial springs, a cooling ring, and a hub made from Inconel 718.
Monolithic ceramic blades made from Kyocera (Somerset, NJ) SN235P silicon nitride.
A small turbine outside diameter (75 mm) is used to lower development cost, but scaling up is possible up to a couple of MW turbines. This sliding-blade ICT prototype is the first of its kind to combine both a functional design and turbine grade materials that are able to withstand a TIT of 1100 °C at a tip speed of 350 m/s.
Experimental Setup
The validation of the performance of the ICT prototype was carried out in a laboratory turbine environment. The ball bearings and compressor wheel are taken out of a Garrett GTX2860R automotive turbocharger unit and retrofit into a custom water-cooled housing. The can combustor is fueled with propane and a hydrogen pilot flame. An array of four K-type thermocouples is placed at the end of the combustion chamber to measure TIT. A remote-actuated bleed valve positioned between the compressor and the combustion chamber is used to control mass-flow. Cooling is provided by an external compressor. The cutaway computer aided design view and a photo of the test rig are showed in Figs. 4 and 5.
Methodology
A two-phase test schedule was carried out to validate the prototype. First, a series of five 5-min tests were run, gradually increasing TIT from 950 to 1100 °C and tip speed from 300 to 350 m/s. Second, an hour-long continuous test was run at 1100 °C TIT and 350 m/s tip speed. During each test, TIT was ramped up to the test temperature in 15 s at a maximum tip speed of 120 m/s (40 krpm). This state was kept constant for 40 s before ramping up to test target speed in 60 s. The test was timed from that point on. At test completion, temperature and rpm were reduced to avoid hot soaking before cutting fuel off. For each test, an average cooling mass flow of 11% of the principal flow was maintained in the cooling ring. This cooling mass flow was used for proof-of-concept stage, but can be largely reduced in future turbines.
Reported TIT are averaged across the four thermocouples placed at the end of the combustion chamber.
Thermochromic paint from Thermal Paint Services is used to measure the temperature of critical turbine parts. Thermochromic paint KN5 is chosen to observe a wide range of temperature (260–1250 °C). Paint color after testing is compared to calibrated coupons. Each coupon is cured as per manufacturer recommendation for high temperature operation (1 h at 260 °C) and then brought up to calibration temperature for 5 min. Calibration coupons from the manufacturer are in accordance with in-house testing and are showed in Fig. 6.
Thermochromic paint (KN5) calibration coupons supplied by Thermal Paint Services (values in °C)
Thermochromic paint (KN5) calibration coupons supplied by Thermal Paint Services (values in °C)
Thermal paint is airbrushed onto all ceramic and metallic components of the turbine wheel.
Results
All tests ran smoothly and no problems or damage occurred. The rpm and TIT plots of the 1 h continuous test are shown in Figs. 7 and 8.
After the 1-h continuous test, visual inspection and disassembly of the prototype revealed no damage. A tested blade is shown in Fig. 9.
Close-up view of silicon nitride (SN235P) blade showing no cracks or damage after hot spin to 1100 °C and 350 m/s tip speed (marks are due to thermochromic paint peeling off)
Close-up view of silicon nitride (SN235P) blade showing no cracks or damage after hot spin to 1100 °C and 350 m/s tip speed (marks are due to thermochromic paint peeling off)
This achievement represents a major step forward in the development of the ICT configuration. Tests are limited in inlet temperature because of the maximum operating temperature of the test rig, but there were no indications of blade or turbine degradation.
Figure 10 shows the temperature observed on the core components of the turbine wheel after testing at 1100 °C for 1 h.
Temperature profile measured with thermochromic paint (indicators A–D refer to Fig. 12)
Temperature profile measured with thermochromic paint (indicators A–D refer to Fig. 12)
No coloration change in the paint occurred in the carbon fiber shroud, meaning this critical component is sufficiently cooled and remains under the curing temperature of the paint (260 °C), therefore well under its glass transition point (335 °C). The cooling ring also remains relatively cold with a maximum temperature of 600 °C at the interface with the blade. Blade temperature ranged 600–900 °C.
Ceramic Blade Life Expectancy Analysis
In order to assess the life expectancy of the tested prototype, thermomechanical finite element (FE) simulations of the turbine are performed.
An FE model is used to analyze temperature distribution and identify critical stress locations in the turbine wheel and determine blade reliability. ansys Workbench is used to carry out a coupled thermal-structural FE analysis. The thermal setup is shown in Fig. 11. TIT used for the analysis is 1100 °C. Surface averaged coefficients of convection applied to the blade profile (840 ) and blade shroud (490 ) are extracted from Ref. [11] while coefficient of convection for the cooling ring (1250 are extracted from a one-dimensional heat transfer model based on theory in Ref. [12]. Since the cooling ring is three-dimensional printed, the effect of high surface roughness due to this manufacturing process is also considered in the calculation of the convection coefficient based on the work of Stimpson et al. [13]. Inlet temperature of the cooling ring is 77 °C, increasing to 150 °C at outlet temperature. In a real engine with a pressure ratio around 8, the cooling ring air supply might need to go through an intercooler. A fixed temperature of 227 °C is applied at the base of the hub and the shaft based on oil temperature measurement made at bearing location. All other surfaces are considered adiabatic. The resultant temperature distribution across the turbine is shown in Fig. 12 and is consistent with temperature measurements obtained with the thermal paint.
Temperature distribution in the turbine predicted by FE analysis with discrete regions where experimental measurements were done
Temperature distribution in the turbine predicted by FE analysis with discrete regions where experimental measurements were done
Thermal analysis results are imported into the structural analysis. Frictionless contacts are used at all metal-to-metal interfaces to ease convergence of the model. A rough contact with radial press fit of 75 μm is used between the cooling ring and the structural shroud. The initial press fit and the high pressure at this interface prevents the cooling ring from slipping relative to the structural shroud. A coefficient of friction of 0.3 is used at the blade (Si3N4) to cooling ring (Inconel 718) interface, based on the work of Sliney and DellaCorte [14]. The resulting maximum principal stresses are shown in Figs. 13 and 14.
for i = 1, 2, 3.
Static fatigue parameters presented in the work of Choi et al. [17] where used. Material constants estimate where provided by Kyocera. The resulting probabilities of a single blade failure (Pf) for different operating times are illustrated in Table 1.
Blade probability of failure and target values
| 1 h | 100 h | 1000 h | 5000 h | |
|---|---|---|---|---|
| Target Pf (estimated range) | 10−7 to 10−8 | 10−5 to 10−6 | 10−4 to 10−5 | 5 × 10−4 to 5 × 10−5 |
| 16 blade core Pf (excluding top 1 mm) | 2 × 10−9 | 1 × 10−7 | 7 × 10−7 | 3 × 10−6 |
| 16 blade set Pf | 3 × 10−5 | 1 × 10−3 | 1 × 10−2 | 6 × 10−2 |
| 1 h | 100 h | 1000 h | 5000 h | |
|---|---|---|---|---|
| Target Pf (estimated range) | 10−7 to 10−8 | 10−5 to 10−6 | 10−4 to 10−5 | 5 × 10−4 to 5 × 10−5 |
| 16 blade core Pf (excluding top 1 mm) | 2 × 10−9 | 1 × 10−7 | 7 × 10−7 | 3 × 10−6 |
| 16 blade set Pf | 3 × 10−5 | 1 × 10−3 | 1 × 10−2 | 6 × 10−2 |
According to the Federal Aviation Administration, multiple turbine engines for small aircraft require a hazardous failure rate of less than 10−7/h [18], which corresponds to a probability of failure of less than 5 × 10−4 for 5000 h of operation if the probability of failure is constant. The corresponding failure probabilities for 16-blade sets for different operating times are shown in Table 1. When excluding the top of the top interface of the blade, the total probability of failure of a 16-blade set is below 10−5 for 5000 h, which meets and even exceeds the target. When including the top of the blade, the failure probability for a full blade set of 16 blades does not currently meet the Federal Aviation Administration targets. The failure probability difference between the core of the blade and the whole blade is caused by the stress profile of the small volume close to the blade to cooling ring interface.
The tensile stress profile in the blade as seen in Fig. 13 clearly shows that stresses are localized at the high-pressure interface between the Inconel cooling ring and the ceramic blade. This is mainly the result of the difference in coefficients of thermal expansion of the two materials. This problem has been documented and solved in the past in a gas turbine application by reducing friction [19]. It is thus plausible to reduce friction and stresses at this interface. The same FE analysis was run at different friction coefficients (μ = 0.1–0.3) to observe its effect on blade Pf after 5000 h of operation. The results are shown in Fig. 15.
cares-calculated failure probability of a single blade after 5000 h of operation versus coefficient of friction of ceramic to metal interface in FE analysis setup
cares-calculated failure probability of a single blade after 5000 h of operation versus coefficient of friction of ceramic to metal interface in FE analysis setup
The trend in numerical results for blade failure in relation to the friction coefficient of the cooling ring to ceramic interface reveals that a small reduction in friction at the interface leads to an exponential drop in failure probability. This indicates that the reduction of friction at this interface could eventually lead to a blade failure probability acceptable for a real engine application.
Conclusion and Perspectives
A sub-MW turbine with typical gas turbine advantages and thermal efficiency over 40% has long been a target for the power market. The ICT configuration is a promising solution to this problem. The introduction of the sliding-blade architecture solved the most important issues encountered by previous ICTs.
This new ICT architecture has led to a prototype that successfully underwent a continuous test of more than 1 h at a TIT of 1100 °C and a tip speed of 350 m/s (115 krpm) without any failure or damage to the turbine.
These achievements mark an important milestone in the development of the ICT, but it is clear that tensile stress at the high-pressure interface at the top of the ceramic blade has to be reduced to achieve acceptable reliability. A numerical sensitivity analysis revealed that the reduction of friction at this interface would lead to an acceptable blade probability of failure. Overcoming this challenge could lead to an engine that opens new possibilities for applications requiring the power density, efficiency and reliability of large gas turbines, but at a smaller scale.
Further design iterations and longer testing of the sliding-blade architecture will be carried out to explore the performance limits of the ICT configuration in terms of tip speed and inlet temperature.
Acknowledgment
The authors would like to express their regards for the technical support provided by the research professionals and professors of Université de Sherbrooke, École de Technologie Supérieure—LAMSI and Exonetik Turbo involved in the development of the ICT.
Funding Data
This work was funded by Exonetik Turbo, the Natural Science and Engineering Research Council of Canada (NSERC) CRD PJ 477320 14 (Funder ID: 10.13039/501100000038), and Defence Research and Development Canada (DRDC) Contract W7714 196710/001/SV (Funder ID: 10.13039/501100002956). Scholarships were provided by the Fonds de recherche du Québec–Nature et technologies (FRQNT).
