Recent/Current Research at the Gas Turbine Laboratory

A Unified Approach for Vaned Diffuser Design in Advanced Centrifugal Compressors

Ruhou Gao
Advisor: Prof. Spakovszky

Modern internal combustion engines require high pressureratio turbochargers and NOx control strategies to improve efficiency and reduce emissions. For such applications, the centrifugal compressor has to simultaneously achieve high efficiency, high pressure ratio and a broad operating range. To meet these requirements, the design trend has been towards highly loaded compressors with high speed impellers and backswept blades for extended operating range and vaned diffusers for enhanced pressure recovery and compact geometry.




The transonic, unsteady and highly non-uniform flow out of the impeller presents multiple challenges for the vaned diffuser. A previous study has identified links between impeller outflow and diffuser performance based on appropriately averaged flow properties. However, the connection between the diffuser design parameters and overall diffuser system performance (which includes the downstream volute) needs yet to be established.


The objectives of the current investigation is to rigorously identify the key geometric parameters that govern diffuser performance and to establish design guidelines for different families of impeller-diffuser combinations. The technical approach combines first principles based modeling with high-fidelity calculations and full scale experiments for performance validation. The goal is to develop a set of unified design guidelines suitable for integration in the preliminary design process.







Characterization and Mitigation of Blade Waviness Effects on Compressor Performance

Jinwook Lee, Albert Gnadt
Advisor: Prof. Spakovszky, Prof. Greitzer

Compressor blade surface waviness results from the manufacturing process. One aspect of the waviness is that it is different than the ‘conventional’, and much-studied, classical boundary layer roughness, which is of much different scale. While there is substantive literature on the effects of surface roughness, transonic flows over wavy wall, transition and the generation of turbulence, there is much less information on the effect of blade surface waviness on compressor performance.

Recent work by Hartmann et al. and Winter et al. shows that a potential loss reduction can be achieved by delaying boundary layer transition. However, the parametric dependence of efficiency on the nature of the waves (non-dimensional amplitude and wavelength) is currently not well defined, and there may be an opportunity to achieve an enhanced performance through achieving tighter tolerances on surface geometry.

This project aims to identify the underlying mechanisms in terms of relevant parameters and sensitivity, and seeks to provide design guidelines for innovative compressor blade profiles robust to surface waviness. The technical approach combines basic experiments in a diffusing passage and numerical analysis where the aerodynamic and geometric parameters can be varied independently.


Investigation of Real Gas Effects in Supercritical CO2 Compression Systems

Derek Paxson, and Dr. Claudio Lettieri
Advisor: Prof. Spakovszky

Reduction of harmful CO2 emissions from power plants is becoming a major concern in industry and a high priority for research, development, and deployment projects. Carbon capture and sequestration in underground wells requires the fluid to be compressed to high pressures (exceeding 100 bar) reaching supercritical conditions. At this state the fluid has density similar to a liquid, but at the same time expends to fill up a volume just like a gas does. Although the properties of these working fluids can be characterized, little is known about their behavior in turbomachinery.

Unlike ideal fluids, thermodynamic properties of real fluids can vary significantly, especially near the critical point, altering the gas dynamics.  The main area of interest in the compressor is the leading edge of the impeller.  Here the flow is accelerated and sees a rapid drop in pressure.  The rapid rate of the expansion leads to non-equilibrium condensation and the fluid entering a metastable state.  This local condensation can lead to a reduction in efficiency and instability in the compressor.  Little is known about the behavior of CO2 in the metastable region, and consequently experimental data is needed to characterize the behavior before it can be modelled in multiphase CFD. 

A laboratory scale high pressure blowdown rig was constructed to assess the behavior of the supercritical CO2. Due to the complexity and cost of constructing and instrumenting a rotating compressor stage, a static test case was needed.  A converging-diverging nozzle with characteristic through-flow times and length scales similar to those of the compressor was chosen as a test case. Nozzles can easily be constructed and instrumented and provide a rapid pressure drop similar to those seen in the compressor.  A rectangular nozzle section with optical access was implemented to allow a shearing interferometer to be used to measure the density of both equilibrium and metastable CO2.  Testing on the rig will be used to evaluate 2-phase models which will be used for CFD of the entire compressor stage geometry.


Example interferometric image of nozzle converging section showing carrier fringe pattern



Investigation of Transcritical Phenomena for Fuel Injection and Mixing Applications

Georgi Subashki , and Dr. Claudio Lettieri
Advisor: Prof. Spakovszky

Inefficient atomization in modern combustors leads to large quantities of unburned fuel which greatly reduces the combustion efficiency. In addition, inhomogeneous concentration of vapor fuel creates regions of higher localized flame temperature, enhancing the formation of harmful pollutants. These challenges can be efficiently addressed by using fuels at supercritical conditions. In particular, recent studies suggest that 80% reduction in NOx emissions and 10% improvement in brake thermal efficiency (Anitescu et al, 2009) can be achieved for injection and fluid mixing at supercritical conditions. In addition, the enhanced mixing of the fuel at supercritical state might enable more compact and lightweight combustor designs.

At supercritical state, fluids exhibit peculiar characteristics: vanishing surface tension, liquid-like density, and gas-like diffusivity. Supercritical fluids can provide potential benefits in a variety of engineering applications but there is little information available in the literature about the transition from subcritical to supercritical state and the related underlying mechanisms. A key challenge is that vanishing surface tension makes standard two-phase modeling inapplicable due to the increased difficulty in defining a phase front. Also, the large variations in fluid properties near the critical point can result in non-ideal, non-equilibrium gas behavior which require accurate and computationally efficient equation of state models.

This project aims to characterize the fluid behavior in transcritical fuel injection and mixing. For the purpose, subcritical and supercritical single-droplet evaporation is studied first. Appropriate non-dimensional groups are determined from a scaling analysis and parametric numerical multi-phase flow calculations are carried out to establish a general fluid-independent model for transcritical phase change. Ultimately, it is desired to integrate such models in the industrial design process for more efficient combustors.


A New Modeling Approach for Rotating Cavitation Instabilities in Rocket Engine Turbopumps

Vincent Wang , and Dr. Claudio Lettieri
Advisor: Prof. Spakovszky

Axial inducers are often used in high performance turbopumps for rocket engines in order to reduce system weight and cost. The inducer allows operation at high rotational speeds and low inlet pressures, which in return can lead to cavitation within the inducer. The dynamic cavitation behavior is often unsteady and periodic, leading to several distinct cavitation instabilities.

Of particular concern among the various instabilities is rotating cavitation, in which cavities propagate circumferentially at frequencies between 1 and 2 times the rotor frequency. In some cases, the propagation frequency can approach the rotor shaft natural frequency, which can cause severe vibrations that present a significant risk to the turbopump. Unfortunately, the physical mechanisms of rotating cavitation are not yet well understood, and as such no general design guidelines exist to mitigate its onset. Existing low-fidelity modeling efforts, which are largely two dimensional, have no predictive capability and fail to capture the three-dimensional nature of the phenomenon.

The goal of this project is to establish a new capability to guide the design of the inducer and related casing treatment to suppress rotating cavitation instabilities. The approach exploits a recently developed body-force based modeling approach that has been used successfully in aerodynamic instability modeling of jet engine axial and centrifugal compressors as well as in acoustic descriptions of non-uniform flow in transonic fans. The feasibility of employing this method to determine the frequencies and mode shapes of cavitation instabilities will be examined. The technical objectives are to (1) define a representative inducer geometry of modern design exhibiting the typically observed dynamic behavior including rotating cavitation, (2) identify the required inducer blade loading distributions and endwall flow forcing requirements to mitigate rotating cavitation, (3) determine the casing treatment type and geometry that can render the required forcing, (4) assess the effect of inlet distortion on cavitation dynamics, and (5) investigate the linkages between rotordynamics and cavitation.



An inducer was designed, manufactured, and experimentally tested. The MIT inducer meets the predicted performance requirements and captures the hypothesized dynamic behavior. At low flow coefficients, cavitation surge is observed while at higher flow coefficients alternate blade cavitation and rotating cavitation are identified.


Modeling Instabilities in High-Pressure Pumping Systems

Dr. Claudio Lettieri
Advisor: Prof. Spakovszky

Pumps are used in the process industry and in power plant applications. They must be able to operate over a wide range of flow rates resulting from variations in load. The particular pumping system under consideration is unstable near its best efficiency point. The challenges are to determine the physical mechanism leading to this system-wide instability and how to redesign the system to ensure stable operation at all relevant flow rates. The system is modeled using a previously-established dynamic modeling framework. This is combined with appropriate boundary conditions to obtain eigenvalues, which are natural frequencies of the system, and their associated growth rates. This determines the stability characteristics of the system. The pumping system model is schematically shown below and includes a plenum containing a gas spring, various area changes, and piping both upstream and downstream of the pump.


Some system components dissipate energy, while others act as energy sources. The pump is an active element, which can dissipate or add energy to the system. The working hypothesis is that the pump is the cause of the observed instabilities. The pump is comprised of an impeller (rotating part) and a volute (stationary part). The impeller dynamics are known. The volute dynamics, however, are not known. It is therefore necessary to obtain a dynamic model of the volute to complete the system stability model. To obtain such a model and to identify the source of the unsteadiness, unsteady RANS CFD calculations of the flow in the volute are conducted. These revealed that the key mechanism governing the self-excited unsteady flow in the volute is bluff body flow separation in the return channel. At the best efficiency operating point, the dynamic behavior in the volute due to this bluff body separation is such that energy is fed into the system. Overall there is insufficient damping, leading to dynamic instability. The figure below illustrates the mechanism which leads to the unsteady behavior in a simplified planar diffuser geometry and relates it to the flow in the volute.

The next step is to conduct a numerical parametric study of the effects of velocity and pressure perturbations on the flow in the volute. This system identification study will result in a dynamic model for use in the existing modeling framework. The framework can then be used to predict at which flow rates unstable operation will occur. Such a prediction tool, coupled with the insight gained into the unsteady flow in the volute, will be used to guide design changes to render the system stable.


Aerodynamic Benefits of Boundary Layer Ingestion for the D8 Double-Bubble Aircraft

Cécile Casses
Advisors: Prof. Greitzer, , Dr. Uranga, Dr. Titchener

Boundary Layer Ingestion (BLI) -- passing a portion of an airframe's wake through the engine -- has been suggested as a means of reducing civil aircraft energy consumption. An aerodynamic performance benefit results from the reduction of viscous dissipation in the airframe wake and in the propulsor jet downstream of the aircraft. A number of challenges exist in the design and performance assessment of BLI aircraft configurations. One is that definition of the propulsion system requirements becomes more difficult because the concepts of thrust and drag, conventionally associated with the engines and airframe, respectively, become ambiguous with a tightly integrated propulsion system. Further, the engine performance itself may be adversely affected by the presence of inlet distortions arising from the ingested airframe boundary layer.

The goal of this research is to determine the aerodynamic BLI benefit of the D8, advanced civil aircraft design (see picture below). The metric used is the mechanical flow power, which is the volume flux of stagnation pressure for the flow regime of interest. Two independent experimental methods have been applied to wind tunnel investigations using a 1:11 scale powered model. The experiments were carried out during two entries in the NASA Langley Research Center 14'x22' Wind Tunnel. The first method is the direct integration of stagnation and static pressure measurements from a five-hole probes at the inlet and exit of the propulsors. This enables information on flow angles, velocity components, and pressure coefficients at the inlet of the propulsors, thereby informing about inlet distortion. The second method assesses the conversion of electrical power into mechanical flow power. To carry out the assessment, experiments to characterize the fans and the electric motors that drive them have been conducted in the MIT GTL 1'x1' Wind Tunnel. Computational assessment is also being carried out by NASA Ames Research Center.

The main results are that the aerodynamic BLI benefit is approximately 8-10%. This comes mainly from the increase in propulsive efficiency (85%) with a small contribution from the decrease in airframe dissipation.


Credit: NASA George Homich

Operability and Engineering Attributes of Artificial Lift System for Oil-Production

Sebastien Mannai
Advisor: Dr. Tan

Oil extraction on many reservoirs requires the use of artificial lift systems to pump the fluid to the surface. While conventional rod pump system work well for vertical well, there is an increasing push towards horizontal wells that is showing the limits of conventional low cost pumping techniques such as rod pump systems. Today there are few effective artificial lift solutions for sweeping fluids through lateral sections. The movements of the rod pump downhole are difficult to evaluate and pumping becomes inefficient and difficult to predict for low flow rate systems.


As such there is a need to assess and define the engineering attributes/requirements of generic low flow rate artificial lift system for an effectivetransport of the oil mixture to the surface for a representative oil production site. The research framework contemplated would include computations, modeling and experiments on generic fluid-pumping system representative of an unconventional fractured horizontal producing well as shown below.


Flow and Heat Transfer in Modern Turbine Rim Seal Cavity

Peter Catalfamo, and Rachel Berg
Advisor: Dr. Tan Sponsor: General Electric

Ingestion of hot gas from the flowpath into the gaps between the rotor and stator can cause turbine components to overheat and lead to deterioration in component life. To prevent this, modern gas turbines, both industrial and aerospace, use compressor bleed air to provide positive outflow through the rim seal (known as “purge” flow). This purge flow can be a substantial fraction of the total flow bled off of the compressor and as such it represents a substantial performance penalty. Past efforts have focused primarily on generating correlative orifice models using experimental data. These results are limited in their applicability by the geometry and conditions tested. In this research MIT, in collaboration with GE Energy and GE Aviation, seeks to investigate the fundamental flow physics in the turbine rim cavity region. Of particular interest is the response of the wheelspace and rim cavity to external stimuli set up by the main annulus flow such as flow unsteadiness due to rotor stator interactions. Rig data being collected by GE will be used to assess the analysis and to guide the investigation. Understanding these mechanisms is fundamental to optimizing seal design and minimizing the purge flow requirements, thus minimizing the associated performance penalty.


Characterization of Performance Limiting Flow Mechanisms in a Centrifugal Compressor Stage

David Erickson
Advisor: Dr. Tan, Sponsor:  GE Aviation

High performance centrifugal compressors featuring vaned or pipe diffusers are commonly utilized in gas turbine engines for today’s small aerospace applications, such as turboprops, rotorcraft, and small business jets.  With advanced aerospace engines trending towards increased overall pressure ratios and decreased core airflow, centrifugal compressors have the potential to see even more widespread use in the future.  However, the physical relationships between centrifugal compressor geometry, the resulting flow behavior, and ultimately the compressor’s performance are not as well understood as they are for axial compressors.  This makes the preliminary design of centrifugal compressors a challenging task, with reliance on empirical models of limited applicability.

This project aims to identify performance limiting flow mechanisms in a centrifugal compressor stage, with a special interest in the diffuser and its influence upstream and downstream, and to describe the causal relationships between these mechanisms and compressor performance (efficiency, pressure rise capability, and operable range).  GE Aviation is providing aerodynamic data measured with high spatial and temporal resolution on a highly instrumented centrifugal compressor utilizing various diffuser geometries and subjected to a wide range of operating parameters.  This data, along with targeted CFD experiments, will be interrogated to identify the flow mechanisms present, and ultimately to explain physically how they give rise to the measured performance trends.



Sensitivity of the D8 Configuration Benefit to changes in Engine Weight, Drag and Efficiency

Yuankang Chen
Advisor: Prof. Greitzer

The D8 jet transport is a conceptual aircraft proposed by MIT in collaboration with industry (Aurora, Pratt & Whitney) and NASA for the 2035 timeframe. The design consists of a 'double-bubble' fuselage, a lifting nose, a pi tail arrangement, and embedded aft engines that ingest part of the fuselage boundary layer. The boundary layer ingestion (BLI) provides a direct fuel saving of roughly 10%, as confirmed in powered model experiments in the NASA Langley 14” x 22 foot Subsonic Wind Tunnel.

There is a need to further quantify the benefit (reduction in fuel burn) of this design over conventional 'tube-and-wing' aircraft, such as the 737, given the technological advances to be expected and, in particular, to determine how sensitive the benefit is to changes in engine weight, drag and efficiency.

Credit: Aurora Flight Sciences

The current research focuses on quantifying the extent of the uncertainties and their impact, i.e. defining how robust the estimated benefits of the D8 configuration are. The objective is to quantify the uncertainty in the D8 benefit, and establish the potential for use in future aircraft designs.


Potential of Emerging Additive Manufacturing Technology on Improving/Enhancing Aerothermal-Mechanical Operating Characteristics of Gas Turbine Engine

Aniwat Tiralap
Advisor: Dr. Tan

The idea and motivation behind this research effort is on exploring the potential of additive manufacturing technology to effectively manage the aerothermal mechanical characteristics of a representative gas turbine engine. It is anticipated that a critical gas turbine flow path where fluid dynamics, heat transfer and structural response all play a role in setting the gas turbine operation will be the focus of this explorative assessment.


Flow Visualization of Compressor Leading Edge Vortex Shedding in Spike-Type Rotating Stall Inception

Andras Kiss
Advisor: Prof. Spakovszky


Compressor rotating stall and surge limit the
stable operating range of gas turbine engines and
can result in dramatically reduced performance
and possibly catastrophic engine failure. The
formation and growth of spike-type stall precursors
is one route to rotating stall that is currently not well
understood. While spike-type stall precursors have
historically been attributed to compressor blade tip
leakage flow, experiments have shown this not to be
the case. Recently, vortex shedding at the rotor tip
leading edge has been hypothesized as the causal
mechanism for spike-type stall precursors.


This work seeks to assess this hypothesis through
a combined numerical and experimental approach.
A linear rotor blade cascade was constructed to
re-create the flowfield of the MIT single stage
compressor during spike-type rotating stall.
Measurements from high-response pressure
transducers installed and a hotwire anemometer
confirmed that the spike precursor was captured
in the cascade environment. Further, smoke
injection at the compressor blade tip visualized
the formation and propagation of the spike-type
stall precursor.  3D URANS calculations of both
the compressor and the cascade, which are in
agreement with the flow visualization results,
confirm leading edge vortex shedding as the
mechanism for spike formation in both


A remaining unknown in spike-type rotating stall
inception is the mechanisms for spike propagation
and growth. Current work is focusing on characterizing
the relative importance of vortex line stretching, viscous
diffusion of vorticity, and the shedding of new vorticity
from adjacent blades on the growth of the spike
precursor. A simplified vortex model is under
development to capture the dynamics of spike
propagation and growth.