While modern engine manufacturers devote significant efforts to the development of reliable and efficient machines, the introduction of novel, optimized components in the hot gas path represents a risky opportunity. Accurate experimental and numerical data are critical to assess the impact of new technologies on the harsh engine environment. The present study addresses the impact of a selection of high-performance rotor blade tips on the aerodynamic and heat flux field of a high pressure turbine (HPT) stage. A combined numerical and experimental approach is employed to characterize the interaction of the tip leakage flow with the rotor secondary flows and the casing heat transfer mechanisms for each individual tip geometry. The turbine stage is tested at engine-scaled conditions in the rotating turbine facility of the von Karman Institute. For the present study, the turbine rotor is operated in rainbow configuration to allow the simultaneous testing of multiple blade tip geometries. RANS simulations are employed to predict the aerodynamic and thermal field of the individual profiles using test-calibrated boundary conditions. Isothermal steady computations are performed at different wall temperatures to compute the adiabatic wall temperature and heat transfer convective coefficient. Low-order models are used to represent the over-tip thermal field and the driving heat transfer mechanisms. The time-resolved outlet flow is characterized using a vortex tracking technique and high frequency aerodynamic measurements to identify the rotor secondary flow structures.

Operating stationary gas turbines on hydrogen-rich fuels offers a pathway to significantly reduce greenhouse gas emissions in the power generation sector. A key challenge in the design of lean-premixed burners, which are flexible in terms of the amount of hydrogen in the fuel across a wide range and still adhere to the required emissions levels, is to prevent flame flashback. However, systematic investigations on flashback at gas turbine relevant conditions to support combustor development are sparse. The current work addresses the need for an improved understanding with an experimental study on boundary layer flashback in a generic swirl burner up to 7.5 bar and 300° C preheat temperature. Methane-hydrogen-air flames with 50 to 85% hydrogen by volume were investigated. High-speed imaging was applied to reveal the flame propagation pathway during flashback events. Flashback limits are reported in terms of the equivalence ratio for a given pressure, preheat temperature, bulk flow velocity and hydrogen content. The wall temperature of the center body along which the flame propagated during flashback events has been controlled by an oil heating/cooling system. This way, the effect any of the control parameters, e.g. pressure, had on the flashback limit was de-coupled from the otherwise inherently associated change in heat load on the wall and thus change in wall temperature. The results show that the preheat temperature has a weaker effect on the flashback propensity than expected. Increasing the pressure from atmospheric conditions to 2.5 bar strongly increases the flashback risk, but hardly affects the flashback limit beyond 2.5 bar.

Wall temperature measurements with fiber coupled online phosphor thermometry were, for the first time, successfully performed in a full scale H-class Siemens gas turbine combustor. Online wall temperatures were obtained during high-pressure combustion tests up to 8 bar at the Siemens CEC test facility. Since optical access to the combustion chamber with fibers being able to provide high laser energies is extremely challenging, we developed a custom-built measurement system, consisting of a water-cooled fiber optic probe and a mobile measurement container. A suitable combination of chemical binder and thermographic phosphor was identified for temperatures up to 1800 K on combustor walls coated with a thermal barrier coating (TBC). To our knowledge these are the first measurements reported with fiber coupled online phosphor thermometry in a full scale high-pressure gas turbine combustor. Details of the setup and the measurement procedures will be presented. The measured signals were influenced by strong background emissions, probably from CO 2 * chemiluminescence. Strategies for correcting background-emissions and data evaluation procedures are discussed. The presented measurement technique enables detailed study of combustor wall temperatures and using this information an optimization of the gas turbine cooling design.

In the second part of this two-part paper, several CHT (Conjugate Heat Transfer) CFD (Computational Fluid Dynamics) simulations were conducted at low temperature and pressure conditions which a 100KW or a 400KW blower can provided, in order to investigate the reason caused the modeling deviation when the matched-Biot number criterion is used for modeling. The CFD method used in this paper is the same as the Part I. Afterwards, several experiments were carried out at a mainstream condition which the 100 KW blower can provide to validate whether the deviations predicted by CFD would appear at that target modeling conditions. The mainstream inlet temperature and pressure are 307 K and 113 kPa, respectively. The model used in this paper is the Mark II turbine as well. The CFD and experiment results show that the main source of the modeling deviation comes from the suction side, the minimum deviation for the four modeling tests has over 0.05 (minimum wall temperature deviation has over 0.75 K). Apart from that, the reason leading to the suction side deviation is the variation of temperature boundary layer, which is caused by the unmatched aerodynamic parameters (such as mainstream Mach number and Reynolds number).

One of the consequences of increasing the efficiency of gas turbine combustors is the higher combustion temperatures within the chamber. Advances on managing larger heat loads have been made to protect the combustor wall and turbines. Among those are thermal barrier coatings (TBCs) deposited on metal walls and forced air cooling such as through effusion holes. Historically, both the flame and TBCs have received a simplified gray treatment using effective absorptivities and emissivities. However, studies have shown that the gray analysis can considerably under-predict the cold metal side temperature resulting in misguided combustor life estimates. In this study, non-gray radiation is compared to gray and black radiation by combining three-dimensional Monte Carlo Ray Tracing (MCRT) solution of non-gray flames in a model gas turbine combustor to one-dimensional energy balance within combustor liners. A recent large eddy simulation (LES) of a gas turbine combustor is analyzed, where both gray and non-gray models are exercised. A two-band spectral model is employed for the TBC, where a translucent band and an opaque band are identified. A line-by-line treatment for gas-phase radiation is adopted, and the incident radiative energy on the combustor wall is collected using the MCRT solver, where the fraction of radiative energy within the translucent band is collected and compared with those obtained from the blackbody assumption. The temperature distributions along the multi-layered combustor wall are computed and parametric comparison is conducted. The effects of the nongray flame radiation are more prominent at elevated pressures than at atmospheric pressure, leading to a difference of approximately 150 K in the prediction of peak temperature on the hot side of the TBC. The gray model is found to over-predict the TBC temperature at downstream locations, but under-predict the TBC temperature near the flame locations. The present study proposes a methodology to estimate the wall temperatures when radiation within the TBC is considered. Future work includes application of the methodology to more realistic combustors where both radiative fluxes and convective fluxes can be accurately captured.

Modern combustor design optimization is contingent on the accurate characterization of the combustor flame side heat loads. Knowledge of regions of high and low heat loads on the liner wall helps designers optimize the cooling designs. The present work focuses on the experimental measurement of the transient heat load along a fused silica (quartz) optical can combustor under reacting conditions for a swirl stabilized premixed methane-air flame. Equivalence ratio was varied from 0.55 to 0.65. Reynolds number based on combustor diameter was varied from 12500 to 18000, where the preheated air temperature was approximately 373 K. The percentage of pilot fuel was varied from 6% to 10% of the main fuel flow rate. Inner and outer walls of the liner were painted with a high temperature flat black paint with an azimuthal offset to aid in infrared measurement of the wall temperature using an infrared camera. Particle Image Velocimetry (PIV) was employed to visualize the flow field for various reacting conditions studied in this work. Based on the heat transfer study, a detailed report of transient heat load along the length of the liner wall for varying reacting conditions has been presented here. The location of impingement of the flame onto the liner and velocity of the flow field were obtained from PIV measurements. Wall heat load at various planes along the length of the liner have been presented. Repeatability of this transient experiment was within 10% between eight different runs for various locations along the length of the liner, except for the region close to flame impingement zone. In the impingement zone, liner heat load varied by about 25% between different runs. It was observed that the change in heat load upstream of the location of impingement on the liner was insignificant with change in pilot ratio as the system tends towards a steady state, contrary to the regions downstream. Higher Reynolds number and equivalence ratios increased the heat load on the liner as expected.

This study includes the design, validation, and fabrication via Direct Metal Laser Sintering (DMLS) of a gas turbine nozzle guide vanes (NGV) that incorporates three innovative cooling schemes specifically enabled by additive manufacturing. The novel NGV design is the culmination of an extensive research and development effort over a period of four years that included low and high speed cascade testing coupled with unsteady CFD for numerous candidate innovative cooling architectures. The final vane design (SJ-vane) consists of sweeping jet (SJ) film cooling holes on the suction surface, sweeping jet impingement holes at the leading edge and double-wall partial length triangular pin-fin with impinging jet at the trailing edge. For comparison purposes, a second DMLS enabled vane (777-vane) was designed and fabricated with prototypical cooling circuits to serve as a baseline. This vane consists of a shaped film cooling holes on the suction surface, circular impingement holes at the leading edge and full length cylindrical pin-fins at the trailing edge. Experiments with the two DMLS enabled vanes were performed at the Ohio State University Turbine Reacting Flow Rig (TuRFR) at engine relevant temperature (1375K) and Mach number conditions. Infrared (IR) thermography was utilized to measure the wall temperature of the pressure and suction surface at several coolant mass flow rates to estimate the overall cooling effectiveness (ϕ). Results showed improved cooling performance for the advanced cooling schemes (sweeping jet film cooling, impingement cooling and triangular pin-fin cooling) compared to the baseline cooling schemes.

For convective heat transfer involving multiple sources of different temperatures in the flow field such as in film cooling, the adiabatic-wall temperature, T ad , is used as the reference temperature to define the heat-transfer coefficient (HTC). Studies based on computational Fluid Dynamics (CFD) have always obtained T ad by requiring the cooled or heated surface to be adiabatic. Similarly, most experimental studies that measured T ad have sought to mimic adiabatic wall by using solids with very low thermal conductivity. Other experimental studies have obtained T ad by making two assumptions: (1) T ad at any given point on the cooled or heated surface is independent of the surface temperature, T s , and the surface heat flux, q s ″ , at that point and (2) the HTC, h ad , which equals q s ″ /( T s – T ad ), depends only on the ratio of q s ″ to T s – T ad . With these two assumptions, measuring q s ″ at two different T s or vice versa at a point yields T ad and h ad at that point. In this study, CFD simulations, based on steady RANS, were performed to assess the assumptions invoked by CFD and experimental studies that seek to obtain T ad and h ad . The assessment was made by studying film cooling of a flat plate with an adiabatic wall and with isothermal walls, where the temperature of the isothermal wall, T s , ranged from the lowest to the highest temperatures in the flow. Results from this study show T ad obtained by enforcing an adiabatic wall does not satisfy the requirement: where and when T s – T ad = 0, q s ″ = 0. Therefore, h ad approaches infinity where T s – T ad is either zero or nearly zero, but q s ″ ≠ 0. Also, obtaining T ad and h ad by measuring two sets of ( T s , q s ″ ) was found to yield non-unique values that depended strongly upon the pair of ( T s , q s ″ ) chosen. To overcome the shortcomings of existing methods, a new method was developed in this study to obtain T ad that does satisfy the requirement: T s – T ad = 0 where and when q s ″ = 0. Also, the method developed yields an h ad that is continuous across T s – T ad = 0. By using the new method developed, errors in T ad and h ad obtained by existing methods were assessed.

As a civil gas turbine cools down, asymmetric natural convective heat transfer causes the bottom sector of the rotor to cool faster than the top; this circumferential thermal gradient can potentially cause the shaft to deflect — a phenomenon called thermal or rotor bow. Rotor bow is tremendously difficult to predict due to its dependence on a number of engine design parameters, in addition to the complex nature of natural convective flows. A novel experimental facility has been developed to gain further understanding into shutdown cooling of a gas turbine. The scope of this paper is to quantify the effect of basic design features on natural convective cooling in an engine annulus during shut-down; specifically the influence of the thermal boundary wall conditions and the annular diameter ratio. In addition to this, a low-cost, robust thermocouple probe has been developed and validated, which allows for accurate temperature measurements in a natural convective boundary layer. An extensive experimental campaign has been completed. The key finding is that the local radial wall temperature difference was found to be the most influential parameter on the local heat transfer. Non-isothermal walls did not alter the overall distribution of the inner wall equivalent conductivity. This was true for both cylindrical and conical sections. An appropriate characteristic length for use in the Rayleigh number definition for both the concentric cylinder and conical sections have been validated. The conical inner section (5° hade angle) did not affect the overall heat transfer in the range of conditions tested. Therefore, the mean surface heat transfer for non-isothermal inner and outer profiles, within the range −0.4 < ΔRa/Ra L c < 0.4, where the thermal gradient is negative in the clockwise from top-dead-centre, can be predicted using isothermal correlations for Ra L c < 5.0 × 10 5 and Dr <= 1.5.

Modern Micro Gas Turbines must be capable to operate at different load points, in order to fulfill the demand of Combined Heat and Power for which they are designed. The combustion chamber structures are therefore subjected to regularly variable thermal loads, yet remaining physically constrained at the rest of the structure. Hence, they experience variable metal temperatures, temperature gradients and thermal stresses which can lead to thermal failure. Typical failure mechanisms in combustion chambers are fatigue and creep. Oxidation can also play an important role. In the present study, Computational Fluid Dynamics methods are used for validation of flow prediction, combustion and heat transfer of an atmospheric combustion chamber. Convective and radiative heat losses towards the ambient are specifically taken into account, leading to better agreement with experimental data from preceding studies. The comparison is presented in this paper. The real-scale machine-operated combustion chamber is then tested at its nominal high-pressure conditions. In this respect, the hereby validated numerical models are employed to simulate the high-pressure operation. A fully coupled Thermo-Structural analysis is performed in order to account for heat fluxes within the solid materials. By so doing, wall temperature distributions can be obtained. The results from Fluid Dynamics simulations serve as input for a Finite Elements Analysis, which provides equivalent stress, strain and deformation distributions by using a linear elastic mechanical model. Such distributions highlight the most critical areas, allowing a first estimate of the components’ life according to thermo-mechanical fatigue. The additional influence of Creep and Oxidation is currently under development at DLR Stuttgart and will be presented in subsequent works.

The high-pressure (HP) turbine and compressor blades are subjected to severe aero-thermal loads either caused by high inlet temperature resulting from the combustion chamber or by the shock wave at the blade surface. In this paper, a partitioned conjugate heat transfer (CHT) approach is used to assess 3D cases with high temperature gradient. On one hand, the high temperature resulting from the combustion chamber generates high azimuthal and radial non uniformities on the nozzle guide vane (NGV), known as hot streak. Since the engine efficiency is directly related to the turbine inlet temperature, manufacturers are seeking to improve thermal barrier coatings (TBC) to allow higher temperature. On the other hand, compressor rotor blades are also subjected to high thermal loads caused by the shock waves. Since the shock wave substantially contributes to the heat transfer, an accurate prediction of the wall temperature is thus required in order to properly estimate the rotor efficiency. However, the thermal load gradient caused by the hot streak, the use of thermal barrier coating or the shock wave have an impact on the overall computation stability. This paper assess the stability for a Dirichlet-Robin interface condition. Indeed an optimal relaxation parameter value is given for the Robin condition to provide a stable CHT computation. The impact on stability of the transient coupling time-step, the coupling frequency and the thermal conductivity is also assessed and the new transient coupling time-step defined allows to correctly stabilize the coupled simulation.

Planar Laser Induced Fluorescence (PLIF) has been demonstrated to investigate a round jet impinging on a flat surface. Detailed thermal field distributions have been obtained near the flat target surface to characterize the wall jet development ensuing from the stagnation point. While PLIF has been demonstrated for combustion applications to measure concentration gradients within a mixture, its application for temperature field measurements is less established. Therefore, the technique was applied to a simple, cylindrical impinging jet. The jet Reynolds number varied with Re jet = 5,000–15,000 while the jet–to–target surface spacing varied from H / D = 4–10. The cooling jet (T jet ∼ 300 K) impinged on a flat, heated surface. The PLIF technique was able to capture the free jet structure and jet development along the target surface. With a short impingement length (H / D = 4), the potential core of the jet strikes the target surface. The thermal gradients captured during the experiments demonstrate the fully turbulent nature of the impinging jet with H / D = 10. The thermal boundary development along the target surface is clearly captured using this fluorescence method. The near wall temperature gradients acquired with the PLIF method have been used to calculate heat transfer coefficients on the heated surface, and these values compare favorably to those measured using a well-established steady state, heat transfer method. The PLIF technique has been demonstrated for this fundamental impingement setup, and it has proven to be applicable to more complex heat transfer and cooling applications.

Aerodynamic cooling of hot components/surfaces such as those encountered in gas turbine engines is needed to avoid premature failure of parts due to thermo-mechanical stresses. An effective way of achieving this cooling is through the exchange of heat via effusion/film cooling holes on the hot surfaces. The gases absorb heat as they flow through the cooling holes and also by forming a protective layer of relatively cool gases near the hot surface. Modeling these processes allow for durable design of components and computational simulations offer a complementary way to design new parts or enable performance assessment of the existing parts at new operating conditions. However, in order to perform numerical simulations of heat transfer through effusion holes, the heat conduction through the solid liner and the convection from the gas phase must be coupled considering all the relevant length and time scales. The time scale separation between the solid and the gas phase makes this prohibitively expensive for large scale computations. In applications involving hundreds of effusion holes, resolving the geometry of each effusion hole along with the primary flow (with typically larger length scales) is very challenging. In the current work, we overcome the difficulties associated with the resolution of cooling holes by employing a local source method (Andreini et al, J. Eng. Gas Turb. Power, 2014) to model the heat transfer to the walls. This method is assessed in a canonical configuration based on experiments performed by Gustafsson (Gustaffson, Ph.D. Thesis, 2001). Large Eddy Simulations (LES) coupled with conjugate heat transfer (CHT) models are used in this study. Simulations that resolve the flow passages explicitly using mesh both in the fluid and the solid domains, were used to validate the fidelity of grid resolution, turbulence models and other simulation parameters in predicting velocity fields and wall temperature data. Although, resolving all the effusions passages provides the most accurate results, it is not practical in real applications. Hence, a local source model is employed to model the heat transfer that happens in the cooling-hole passages. In this method, the effusion passage is not resolved (using a mesh) and the mass transfer across the cooling hole passage is prescribed as an injection-extraction boundary condition. The heat transfer at the fluid-solid interface of the cooling-hole passage is also modeled based on Nusselt number correlations available in the literature. This modeling procedure enables simulations with flexible mesh topologies that can be generated at a relatively low cost in comparison to the fully resolved mesh configurations. The local source method is assessed and validated using the available experimental data. The results show that the meshes which resolve the penetration depth in the solid and which conform at the solid-gas interface provide better prediction of the wall temperatures.

Previous researches on latticework were focused on the convective heat transfer performance on pressure and suction sides of a blade model. Besides, it has an effect on leading edge by impingement. Thus, the present study provides heat transfer enhancement and pressure loss of jet impingement of a latticework on side wall in a wide channel (AR = 4). Two latticework configurations with impingement effects are employed in this study. Three kinds of sub-channel models are used in this experiment, which is according to different cooling designs. The angle of the rib is 45° and the numbers of subchannel are 4, 6 and 8, respectively. Reynolds number range is from 10000 to 30000 with an increment of 10000. The wall temperature is obtained by using wide band liquid crystal technique, and then the heat transfer coefficients on the target surface of the channel are achieved. Pressure drop of the latticework channel is also measured by pressure taps. The result shows that these two latticework models have different flow and heat transfer characteristics. The Nusselt number distribution is not similar to that of traditional jet array impingement. The range of Nusselt number enhancement is 2.3 to 6.4 compared to that of a smooth convective channel (the Nusselt number is based on the channel hydraulic diameter). The jet-to-target distance could reduce the overall averaged heat transfer on side wall. But it could also lead to a high Nu region. To the normal lattice model, more sub-channels there are, more pressure loss it has. To the novel lattice model, sub-channel number cannot affect the pressure loss, but the jet-to-target distance could affect the friction factor obviously.

In the present paper, lattice geometries have been developed and tested for application in internal cooling of gas turbine blades. These geometries are suitable for embedding within a near surface or double wall cooling configuration. Two types of unit cells with variation in number of ligaments and porosity, were used to generate two lattice configurations. The first type of unit cell consisted of six ligaments of 0.5 mm diameter joined at a common vertex situated at the middle. As such, three ligaments were located on each side of the common middle vertex. In addition, the top half of the unit cell was rotated 60 degrees compared to the bottom half of the unit cell. The second type of unit cell was derived from the first type but with four mutually perpendicular ligaments added in the middle plane. Two lattices, referred to as L1 and L2, were obtained by repeating the type 1 and type 2 unit cells, respectively, in both streamwise and spanwise directions. The test coupons consisted of these lattice structures embedded inside a channel of 2.54 mm height, 38.07 mm width and 38.1 mm in length. The coupons were fabricated using Inconel 718 powder through selective laser sintering (SLS) process. The heat transfer and pressure drop performance was evaluated using steady state tests with constant wall temperature boundary condition and for channel Reynolds number ranging from 2,800 to 15,000. In addition, steady state numerical conjugate heat transfer simulations were conducted to obtain a detailed insight into the prevalent flow field and temperature distribution. Experimental results showed a heat transfer enhancement of upto 3 times at the highest Reynolds number compared to a smooth channel. Both the heat transfer and pressure drop increased with a decrease in the porosity from L1 to L2. The numerical results revealed formation of prominent vortical structures in the inter-unit cell spaces. These vortical structures were located in the upper half of the channel causing an increased heat transfer on the top end wall. As a result, these lattice structures provided an augmented heat transfer with a potential for favorable redistribution of the coolant.

The cooling ratio on a dynamically forced 7×7 impingement jet array is studied experimentally. The current study is focused on determining the influence of a phase-shift between every row of nozzles as well as the impact of a duty cycle variation on the cooling ratio. Both parameters are studied in dependency of the impingement distance (H/D = 2, 3, 5), the (nozzle-)Reynolds-number (Re D = 3200, 5200, 7200) and the excitation frequency (f = 0 Hz – 1000 Hz). For every set of parameters, the phase-shift between every row of nozzles is varied between Φ = 0% – 90% while the variation of the duty-cycle is performed between DC = 35% – 65%. During all investigations, the dimensionless distance between adjacent nozzles is fixed at x/D = y/D = 5 and liquid crystal thermography is used to acquire the wall temperatures, which are further processed to calculate the local Nusselt numbers. Generally, the implementation of an excitation frequency allows a case depending increase in cooling ratio of up to 52%. Further implementation of a phase-shift yields an additional frequency-depending improvement of the cooling ratio. In case of duty cycle variation, the best case revealed an additional 19% improvement in cooling ratio.

The supersonic outlet conditions from a rotating detonation combustor exhibit fluctuations in temperature and pressure that exceed 200% of their mean level. Such unsteady conditions will induce a large convective heat loading onto a downstream supersonic turbine. Hence, the precise evaluation of the thermal load to the vane and rotor is essential to the design of adequate cooling strategies. In this paper, a numerical framework is proposed to compute the convective heat transfer on two types of supersonic turbines: axial and radial outflow. The fluctuations imposed at the turbine inlet were obtained from a nozzle coupled to a rotating detonation combustor. Both radial and axial turbines were designed and subsequently analyzed with full stage unsteady simulations using an Unsteady Reynolds Averaged Navier–Stokes solver. The inlet boundary conditions to the turbine are based on CFD results from a rotating detonation combustor. The unsteady adiabatic convective heat transfer coefficient was obtained from two simulations performed at a fixed homogeneous wall temperature. The heat flux variation in span-wise and stream-wise direction is analyzed in detail. Budgeting of the unsteady heat flux mechanism was performed to identify the driving contributor of the heat transfer within the turbine and finally both designs are compared.

The process of ignition in aero-engines raises many practical issues that need to be faced during the design process. Recent experiments and simulations have provided detailed insights on ignition in single-injector configurations and on the light-round sequence in annular combustors. It was shown that Large Eddy Simulation (LES) was able to reliably reproduce the physical phenomena involved in the ignition of both perfectly premixed and liquid spray flames. The present study aims at further extending the knowledge on flame propagation during the ignition of annular multiple injector combustors by focusing the attention on the effects of heat losses, which have not been accounted for in numerical calculations before. This problem is examined by developing Large Eddy Simulations of the light-round process with a fixed temperature at the solid boundaries. Calculations are carried out for a laboratory-scale annular system. Results are compared in terms of flame shape and light-round duration with available experiments and with an adiabatic LES serving as a reference. Wall heat losses lead to a significant reduction in the flame propagation velocity as observed experimentally. However, the LES underestimates this effect and leads to a globally shorter light-round. To better understand this discrepancy, the study focuses then on the analysis of the near wall region where the velocity and temperature boundary layers must be carefully described. An a-priori analysis underlines the shortcomings associated to the chosen wall law by considering a more advanced wall model that fully accounts for variable thermophysical properties and for the unsteadiness of the boundary layer.

This paper presents a numerical investigation of natural gas combustion in an OPRA OP16 tubular DLE combustor operating in premixed mode. The aim of this study is to compare different CFD models for the accurate prediction of emissions. The numerical results are validated against experimental results obtained in an atmospheric combustion test rig. A CFD workflow using AVL FIRE software has been devised enabling accurate CO and NO x emission calculations. Different turbulence models have been employed for adiabatic flow simulations at nominal load: the k-ζ-f turbulence model, PANS and LES-CSM model. Two detailed reaction mechanisms have been used to model combustion chemistry and thermodynamic properties: the DRM19 mechanism and the GRI 3.0 reaction mechanism. In the adiabatic simulations both the k-ζ-f and LES-CSM turbulence models perform well in combination with the GRI 3.0 reaction mechanism. In the non-adiabatic simulations heat exchange between combustor liner and cooling air is taken into account by using a wall temperature profile derived by scaling the experimentally obtained temperature profile such that the requirement of conservation of energy is satisfied. Reactive flow simulations at nominal load using the k-ζ-f turbulence model and GRI 3.0 reaction mechanism resulted in accurate predictions for time-averaged combustor pressure drop, CO and NO emissions: the predicted CO value deviates less than 20% from its experimentally obtained value, for NO the deviation is less than 10%. Inclusion of heat exchange between combustor and cooling air liner reduces this deviation to less than 10% for CO and less than 2% for NO. At a load equal to 106% of its nominal value, non-adiabatic simulations result in less satisfactory CO and NO predictions: NO is over predicted by approx. 20% while CO is overpredicted by nearly a factor of four. The latter is caused by the fact that experimentally obtained CO emissions reduce significantly when the load is increased from 100% to 106%, while numerically obtained CO emissions just change by a few ppm.

Time-accurate 3-D CFD simulations based on the SST-SAS turbulence model were performed to study the effects of heat load on the unsteady flow and heat transfer in a cooling duct with a staggered array of short pin fins. For this duct, the static pressure at its exit is maintained at 25 bars, and the cooling air that enters has a temperature of 673 K with a flow rate that produces a Reynolds number of 25,000. To examine the effects of heat load, the following isothermal wall temperatures were studied: 678 K, 873 K, 1073 K, and 1,273 K, which give rises to heat loads that range from 15 kW/m 2 to 1.5 MW/m 2 . Results obtained show high heat loads to cause considerable changes in the temperature of the cooling flow along the duct, which causes significant changes in density and velocity as well as viscosity and thermal conductivity. These changes along the duct were found to affect the locations where unsteady flow separation take place around the pin fins, the magnitude of the vorticity shed in the wakes, and the shedding Strouhal number. These unsteady flow mechanisms in turn strongly affect the nature of the surface heat transfer. A correlation formula for the heat transfer, which accounts for the effects of heat loads, was developed.