The reliability of temperature probes in gas turbines is dependent on their capability to withstand the harsh environment they experience when installed in an engine. The severe conditions sensors experience during operation requires the use of appropriate methodologies to assess their life. During the ALPHET project the University of Cambridge and Esterline Advanced Sensors have been working on developing an advanced lifing methodology: in order to predict the life of thermocouple probes used in gas turbines, this combines accelerated laboratory tests based on the salt deposition method, combustor rig tests, metallurgical analysis of probes returned after engine service and information extracted from their associated temperature data acquired during their long term operation in aero gas turbines. The work included analysis of materials used in current temperature probes and investigation of advanced materials for future thermocouple sensors that will work at temperatures higher than today’s probes. This paper will summarise the main features of the improved lifing methodology and the challenges associated with the use of complex engineering data obtained from engines.

This paper presents a numerical investigation on the rationality and reliability of film cooling experiments, in which the realistic temperature ratio (TR) of mainstream to cooling air are substituted by the density ratio (DR) of ambient air to a foreign gas with high density. Today, advanced gas turbines operate at much higher temperatures than the allowable temperature of turbine component materials, which makes it very difficult to achieve real TR in the most of laboratory environments, because the real TRs are usually larger than 2.0. Therefore, the foreign gas (for example CO 2 ) with high density was widely used to obtain a proper density ratio for simulating the mixing process of cooling air with high temperature mainstream thereby. However, the TR effect on film cooling is not completely replaced by DR, because the influence factors of film cooling performances are not only DR, but also the thermal properties of cooling gas and mainstream, such as specific heat capacity, viscosity and conductivity. In this work, a film-cooled endwall is used as specimen, and DRs are controlled by two ways, i.e. changing mainstream temperatures, and using a foreign gas, respectively. The numerical results of film cooling performances at the same DR obtained by the two ways are analyzed and compared. The analysis reveals the difference of film cooling performances between TR and DR on the film-cooled endwall, and the comparison indicated that the errors caused by the substitution of foreign gas are acceptable, only when BR, DR, Re ∞ and T ∞ are all small, but when BR, DR or Re ∞ increases, the relative error cannot be neglected, and it may reach 30% in real running conditions.

Wet compression is a strategy adopted to increase the power output of gas turbines, with respect to dry conditions, usually also incrementing the operating range of the compressor. However, stall and surge are two aerodynamic instabilities which depend on many factors, and they are expected to occur even in wet compression at low flow rates. Despite the many studies carried out in the last 80 years, literature does not offer many works concerning these instability phenomena in wet compression. In this paper, an experimental analysis of stall and surge in wet compression conditions is carried out on an axial-centrifugal compressor installed in an existing test rig at the Engineering Department of the University of Ferrara. Some modifications of the test rig were necessary. The intake duct was implemented with a water injection system which, by means of water spray injectors, allows the uniform mixing of air and water before the compressor inlet. The control and data acquisition system of the test bench was updated with new hardware and software to obtain faster data sampling. Transient and steady-state tests were carried out to make a comparison with the experimental results in dry conditions. The analysis was carried out using traditional thermodynamic sensors, by means of both classic post-processing techniques, and cyclostationary analysis. The aim is to i) evaluate the influence of wet compression on the stable performance of the compressor ii) qualitatively identify the characteristics of stall and surge in wet compression by means of sensors which were shown to capture these phenomena well and iii) demonstrate the reliability of cyclostationary analysis in wet compression conditions for stall and surge analysis.

The air handling system for large diesel/gas engines such as those used on locomotive, marine, and power generation applications require turbochargers with a high reliability and with turbomachinery capable to adjust to different operating conditions and transient requirements. The usage of variable geometry turbocharging (VGT) provides flexibility to the air handling system but adds complexity, cost and reduces the reliability of the turbocharger in exchange for improved engine performance and transient response. For this reason, it was desirable to explore designs that could provide the variability required by the air handling system, without the efficiency penalty of a conventional waste gate and with as little added complexity as possible. The current work describes a new low-cost variable geometry turbine design to address these requirements. The new tandem nozzle concept proposed is applicable to both axial and radial turbines, and has been designed using conventional 1D models and 2D/3D CFD methods. The concept has furthermore been validated experimentally on two different test rigs. In order to avoid the long lead times of procuring castings, the nozzle for the axial turbine was manufactured using new additive manufacturing techniques. Both the axial turbine and the radial turbine designs showed that the concept is capable to achieve a mass flow variability of more than 15% and provide a robust and cost-effective alternative to conventional VGT designs by significantly reducing the number of moveable parts.

The basic principle of a distinct idea to reduce an aerodynamic mixing loss induced by the difference in tangential velocity between mainstream flow and rotor shroud leakage flow is presented in “Part I – Design Concept and Typical Performance of a Swirl Breaker” The design concept offers an effective geometry for improving steam turbine stage efficiency. When the swirl breaker is installed in the circulating region of leakage flow at the rotor shroud exit cavity, the axial distance between the swirl breaker and rotor shroud is a crucial factor to trap the leakage flow into the swirl breaker cavity. In this Part II of the study, five cases of swirl breaker geometry with different axial distances between the swirl breaker and rotor shroud, which covered a range for the stage axial distance of actual high and intermediate (HIP) pressure steam turbines, were investigated using computational fluid dynamics (CFD) analysis and tests. Compared to a conventional single-stage CFD analysis, by conducting an additional single-rotor analysis with the modified shear stress transport (SST) model coefficient, the prediction accuracy for typical improvements in stage efficiency was increased in comparison to the single-stage analysis with the default SST model. Based on CFD results, the verification tests were conducted in a 1.5-stage air model turbine. By decreasing the axial distance between the swirl breaker and rotor shroud, the tangential velocity and the mixing region in the tip side which is influenced by the rotor shroud leakage flow were decreased and the stage efficiency was increased. The case of the shortest axial distance between the swirl breaker and rotor shroud increased turbine stage efficiency by 0.7% compared to the conventional cavity geometry. In addition, the unsteady pressure was measured in the swirl breaker cavity to evaluate the structural reliability of the swirl breaker. These results showed the maximum pressure fluctuation was only 0.7% of the entire flow pressure. Consequently, both performance characteristics and structural reliability of swirl breaker were verified for application to real steam turbines.

While reliability stays the most critical requirement for power generation in the Oil & Gas sector, energy effectiveness is becoming an increasingly important topic. Gas turbines are efficient and flexible tools both in electrical and mechanical drive applications as they cope with the multiple energy demand profiles of this sector. Aeroderivative gas turbines boast fast response and fast O&M procedures which fits with the exploitation of hydrocarbon fields. Heavy duty machines are robust and very popular in refinery utilities or as large mechanical drivers. However, it is sometimes desirable to boost GT performances: operators of the Oil & Gas sector may need to increase power output or wish to improve efficiency, especially during part load operation where the latter is degraded. While a number of commercial hardware, service and software products cover these needs, the operators may want to keep their own control on their production tools. To that end resorting to simple thermodynamic considerations enables rule-of-thumb evaluations of the impact of cycle changes on GT performances but this possibility is often unexploited. This joint paper is intended to provide the energy community, especially the engineers of the Oil & Gas sector, with some basic methods for the estimation/anticipation of gas turbine performance changes.

Industrial steam turbine applications require a high flexibility in terms of aerodynamically as well as mechanically operating conditions. Mechanical drives run at variable rotor speeds and often face high backpressure levels, which is a particularly challenging aspect for the last-stage moving blade (LSMB) row of a low-pressure (LP) steam turbine. As speed synchronous blade resonances cannot be excluded over the entire operating speed range, the LSMB needs to be designed resonance proof from viewpoint of dynamics and mechanical strength. The subsequent paper describes a new developed LP stage group for variable speed applications, which is available as fully scaled blade family with distinct exhaust area and rotational speed limit reaching from conventional 3000rpm up to high speed utilities for special purposes. It consists of two standard stages, which can be operated at off resonance as well as resonance conditions. For that reason frictional elements were implemented in both moving blade rows. They are loosely assembled into pockets, which are placed on the pressure as well as suction side of each airfoil. Basically, this feature was already introduced for certain Siemens LP blades, see [1], [2], [3] and [4]. However, the recent development on this field utilizes several design changes for the benefit of robustness and reliability. The present paper focuses on the non-linear dynamic behavior of the friction damped, coupled blade rows and outlines the underlying design method as well as calculation process.

Blade fatigue fracture is the main factor affecting the safe operation of the turbine and the aerodynamic exciting force caused by partial admission accelerates the fatigue failure of the control stage blade. In this paper, based on the three-dimensional unsteady flow research of 300MW control stage, a new type of inlet chamber structure is proposed, which can reduce the local aerodynamic exciting force effectively. The performance of original structure and new structure is compared at 0.7 partial admission operation condition in details. Firstly, the paper gives the overall performance of the control stage under the two structures. It is proved that the new inlet chamber structure proposed in this paper only has a slight influence on the overall performance, which is shown as mass flow, power and efficiency. Secondly, distributions of flow parameters such as pressure, static entropy and axial velocity at the stator inlet and rotor inlet are given in the paper. The new structure is found to alleviate the sudden change of flow parameters such as pressure at the inlet steam chamber transition region. Finally, five key points of axial aerodynamic force and seven key points of tangential aerodynamic force are captured to illustrate the influence of partial admission on instantaneous aerodynamic exciting force. Further, the mechanism of new inlet chamber structure on local aerodynamic exciting forces is explained. The result shows that the new inlet chamber structure can reduce the local axial aerodynamic exciting force up to 13% and reduce the local tangential aerodynamic exciting force up to 29.7%, which is beneficial to improve the safety and reliability of the turbine operation.

Considering the increasing development of naval destroyers and the obvious advantages of marine gas turbine, this paper designs a novel COGAG (Combined Gas-turbine And Gas-turbine) propulsion system which mainly consists of four GT25 marine gas turbines and one CCG (Cross Connection Gears) for the large destroyer. Firstly, the overall configuration and key devices of COGAG propulsion system are introduced briefly. Then, the typical operating patterns of COGAG propulsion system under different condition are discussed in detail. Finally, many experimental information are further presented. All the results show that the developed COGAG propulsion system not only has higher flexibility and reliability, but also can effectively improve the fuel economy of marine ships.

The leakage flows within the gap between the tips of unshrouded rotor blades and the stationary casing of high-speed turbines are the source of significant aerodynamic losses and thermal stresses. In the pursuit for higher component performance and reliability, shaping the tip geometry offers a considerable potential to modulate the rotor tip flows and to weaken the heat transfer onto the blade and casing. Nevertheless, a critical shortage of combined experimental and numerical studies addressing the flow and loss generation mechanisms of advanced tip profiles persists in the open literature. A comprehensive study is presented in this two-part paper that investigates the influence of blade tip geometry on the aerother-modynamics of a high-speed turbine. An experimental and numerical campaign has been performed on a high-pressure turbine stage adopting three different blade tip profiles. The aerothermal performance of two optimized tip geometries (one with a full three-dimensional contoured shape and the other featuring a multi-cavity squealer-like tip) is compared against that of a regular squealer geometry. In the second part of this paper, we report a detailed analysis on the aerodynamics of the turbine as a function of the blade tip geometry. Reynolds-averaged Navier-Stokes simulations, adopting the Spalart-Allmaras turbulence model and experimental boundary conditions, were run on high-density unstructured meshes using the Numeca FINE/Open solver. The simulations were validated against time-averaged and time-resolved experimental data collected in an instrumented turbine stage specifically set up for the simultaneous testing of multiple blade tips at scaled engine-representative conditions. The tip flow physics is explored to explain variations in turbine performance as a function of the tip geometry. Denton’s mixing loss model is applied to the predicted tip gap aerodynamic field to identify and quantify the loss reduction mechanisms of the alternative tip designs. An advanced method based on the local triple decomposition of relative motion is used to track the location, size and intensity of the vortical flow structures arising from the interaction between the tip leakage flow and the main gas path. Ultimately, the comparison between the unconventional tip profiles and the baseline squealer tip highlights distinct aerodynamic features in the associated gap flow field. The flow analysis provides guidelines for the designer to assess the impact of specific tip design strategies on the turbine aerodynamics and rotor heat transfer.

Axial fans are widely used in modern industry and new regulations and stringent environmental concerns are prompting manufacturer to design efficient low-noise axial fans. This paper is focused on improving the aerodynamic performances and reducing the tonal noise at BPF and its harmonics by the optimum choice of lean-swept blade and the stacking line for the low-speed axial fan. The aerodynamic characteristics of the axial fan with a shroud are explored by CFD with ANASYS CFX. A hybrid method, SST turbulence model for flow and FW-H equation for acoustics, is chosen to predict the radiated noise. The accuracy and reliability of predicted aerodynamic and aeroacoustics results are verified by comparing both computation and experimental data. A number of modified blades with different leaned angle, swept angle and the stacking lines are modeled and analyzed, and the investigation into the optimum choice of lean-swept blade and the stacking line is conducted according to aerodynamic performances and tonal noise. Q-criterion which can visualize the major flow disturbances is applied for the purpose of identification of acoustic sources. The turbulent flow structures on the leading edge, tip and suction side of the blade are main noise sources. An optimal modification is determined through the analysis of the aerodynamic performances and noise, which is to achieve the desired performances by blade sweep and lean and adjusting the stacking line. The results show that aerodynamic and acoustic performances of the optimized fan are better than that of the original fan and the improvement is more obvious to change the stacking line with centre of gravity compare to blade sweep and lean for the low-speed axial fan.

The development of a 60-kWe turbo generator that uses supercritical carbon dioxide (sCO 2 ) cycle technology at the lab scale is described herein. The design concept for the turbo generator involved using commercially available components to reduce the developmental time and to increase the reliability of the machine. The developed supercritical partial-admission CO 2 turbine has a single-stage axial-type design with a 73-mm rotor mean diameter. The design of the sCO 2 turbine uses impulse and partial admission to reduce the axial force and rotational speed. We simulated the flow of the designed sCO 2 turbine. To increase the simulation accuracy, a real gas property table is coupled with the flow solver. The turbine performance test apparatus and test results are described; then, the turbine is continuously operated for 44 min. The maximum turbine power is 25.4 kW, and the maximum electric power is 10.3 kWe.

High-speed train is developing popular in China, which provides the convenient and fast transportation way, comparable to plane. The moving direction and speed of high-speed train is decided by the traction motor. Generally, a coaxial centrifugal fan is used to cool the motor and assemble in the motor casing. To ensure the reliability of the traction motor, more and more attention is paid to improve the performance of cooling fans in a wide range of rotating speed. As the train is designed to move in both directions, the traction motor is designed to rotate in both directions, so does the coaxial motor cooling fan. Symmetrical and straight blade structure is adopted to get the same performance of the fan in both forward and reverse moving directions. Therefore, the aerodynamic performance of the cooling fan is relatively not good enough, which results in relatively high aerodynamic noise. In order to analyze the cooling fan aerodynamic performance and aerodynamic noise, CFD method was performed on the full 3D model with the impeller-casing clearance. The acoustic analogy method was used to analyze the noise of the centrifugal cooling fan. In addition, the aerodynamic noise of the motor with the cooling fan was tested at different rotating speed in the semi-anechoic lab. The CFD method is verified and the results are in good agreement with the experimental results. The results show that it is necessary to consider the effects of impeller-casing leakage and the vacuum inlet condition in the simulated model to get its more accurate performance. Modified CFD model of the cooling fan was proposed here. It is suggested that the modified structure of the casing can be used to improve the performance of the cooling fan and reduce the corresponding aerodynamic noise.

Efficiency, operational flexibility, durability, environmental friendliness, and reliability — these words are often used to define expectations and requirements for product performance, required in the energy market today. Yet, depending on where one looks at the curve of product evolution, these words have very different meanings. One thing remains constant, irrespective of where one looks at performance growth over time, product improvement requires actual field experience data on a number of operating plants and units, to supplement the testing and verification processes of the Original Equipment Manufacturer (OEM). The combination of the field experience with the latest testing, verification, and quality control techniques, followed by the OEM, is essential to achieve the inherent reliability and ultimately, the life-time expectations set by the market. There is always a question as to the amount of actual field experience data that is necessary to validate product improvement or advancement. Today, the market seems to accept a lower standard of operating time (∼ 8,000 Actual Operating Hours (AOH)), compared with the past (∼ 100,000 AOH) for product validation. If Equivalent Operating Hours (EOH) were the basis for assessing the maturity of a new technology, the times would be lower. Whether the number is 8,000 or 100,000 AOH, the fact is that field experience data represents the intersection of the OEM’s design, validated through component and system level testing, coupled with the actual operating and maintenance practice and experience of the owner/operator. It is here that the OEM and owner/operator begin to directly share a common objective; the ability to demonstrate the efficiency, operational flexibility, durability, environmental friendliness and reliability of the product improvement. Consequently, the OEM and the owner/operator have a strong and shared interest in the accuracy, fidelity, and the completeness of field data; to ensure that an effective Failure Reporting and Corrective Action (FRACA) process is in place — driving continuous product improvement; to meet market expectations for equipment performance, and to support new product introduction. The focus of this paper is field experience data, past and present, with a specific emphasis on the increasing influence and value that data has in the ever changing and competitive energy market. The paper will suggest that the present and future need for high fidelity equipment data (at a component level of detail) is not just essential for supporting engineering efforts for product evolution, but also to support effective Operations & Maintenance (O&M) strategies. The paper will advance the notion that the fusion of total plant data, from three primary sources, with the ability to transform, analyze, and act based on integrating subject matter expertise is essential for effectively managing assets for optimum performance and profitability; executing and delivering on the promise of “Big Data” and advanced analytics.

Since the unsteadiness of tip leakage flow has profound effects on both aerodynamic performance and stall margin of axial compressors, it is important to accurately predict the transient tip flow at affordable computational cost. Limited by the high requirement of grid resolution of wall turbulence flow, large eddy simulation (LES) method is greatly restricted in engineering application. In the present work, a Reynolds-stress-constrained large eddy simulation (CLES) method has been introduced, in which the whole domain is simulated using LES while Reynolds stress constraint is enforced on the subgrid-scale (SGS) stress model for near-wall regions aiming at reducing the near-wall grid resolution. The CLES simulations have been performed to investigate the flow behaviors of the unsteady tip leakage flow in a transonic compressor NASA Rotor 67 at near-stall conditions. Reliability assessments have been conducted through comparisons of experimental measurements and numerical results obtained by RANS, DES, CLES as well as LES, respectively. Both the total pressure ratio and isentropic efficiency calculated by CLES agree well with experiment. The turbulence statistical results show three distinct high flow fluctuation regions near the blade tip. The first one is a long and narrow strip ahead of the leading edge of the rotor caused by the movement of the passage shock wave. The second one is formed on the suction side from the leading-edge of the rotor blade due to the oscillation of the tip leakage vortex. And the third one, which occupies most of the blade passage from the middle part of the rotor blade, is generated under multiple factors. The frequency characteristic of the unsteady tip leakage flow has been analyzed. The energy spectrums of the local transient pressure signals are highly related with the local unsteady flow features. The originating mechanisms of the flow unsteadiness in the rotor tip leakage flow have also been discussed, and the results show that the flow unsteadiness is mainly caused by a combined interaction effect of the double leakage flow, the tip leakage vortex flow spilled from the adjacent blade passage, as well as the involved main flow.

In the current economic and political environment, there is a push for gas turbine operators to achieve higher operating efficiencies, which in turn, reduces emissions and fuel consumption. As these owners and operators seek to increase the efficiency of their machines, they are increasingly turning to physics-based performance modeling. This allows the end user to analyze machine performance, plan for performance upgrades, and evaluate use cases and operating conditions not originally envisioned by the original equipment manufacturers (OEMs). For owners/operators who do not have access to physics-based models provided by the hardware OEM or would like to evaluate modifications to legacy hardware, physics-based models may be developed using measured turbine performance data and high-level knowledge of the turbine architecture. In previous work, a physics-based performance model of an industrial gas turbine engine was created using measured plant operating data and an inverse solution method to allow off-design exploration of its performance. However, this model’s uncertainty was unknown, and knowledge of uncertainty is crucial to understanding a model’s reliability. In the present work, the model’s uncertainty in predicted performance at a particular operating point is investigated using statistical methods. Polynomial regressions of standard deviation are used alongside the performance regressions to describe the uncertainty at various operating points. These regressions are also used to visualize the variation of uncertainty across the performance map. Such knowledge of uncertainty can aid gas turbine operators in decision making with regard to the risks of off-design operation or equipment modifications.

The development of gas turbines, Mitsubishi Hitachi Power Systems, Ltd. (MHPS) has continued to pursue and contribute to society in terms of global environmental conservation and stable energy supply. MHPS leverages its abundant gas turbine operation experience and takes advantage of its extensive advanced technologies research on the Japanese National Project. MHPS has been participating in this project since 2004. Recent years’ achievements include the demonstration of a gas turbine combined cycle (GTCC) efficiency in excess of 62% created by increasing the turbine inlet temperature to the 1,600°C class in the M501J in 2011. The Latest M701F incorporates “J” gas turbine technologies, already applied to actual equipment, for efficiency improvement. It also applies air-cooled combustor technologies successfully validated in the G class, for increased flexibility. The 1st unit started commercial operation in 2015 and currently 4 units has accumulated more than 46,000 actual operating hours collectively. MHPS is making the upgrading program for existing F-series gas turbines. The proven technology verified in the M501J and developed in the National project increases efficiency and reliability. This paper explains the features and development status of Latest M701F gas turbine, and explains upgrade program for existing F-series gas turbines.

As the first core component of commercial aviation engines, compressor is likely to ingest a serious amount of liquid water into its path flow during the situation of heavy rain. Under the influence of centrifugal force, the tip region of compressor rotors may be filled with water. Unfortunately, the instability of tip clearance flow is a major reason for rotating stall. In order to explore the effects of water ingestion on the steady operating boundary of compressor, the change characteristics of tip clearance flow have been investigated in this research by using numerical simulation method. In reality, the compression process after water ingestion can be regarded as a special wet compression. Therefore, on the basis of existing wet compression study, the ingested mass flow and the mean droplet diameters were set as 1% of inlet air flow as well as 100 μm respectively to investigate how does water ingestion affects the flow field at the tip clearance of rotors. The typical trajectory of water droplet in compressor has been researched firstly to simply understand the situation of water ingestion and to verify the reliability of some special droplet breakup models. Then, the compressor performance in two states of dry and wet have been compared to indicate that the pressure ratio and adiabatic efficiency of compressor will decrease after water ingestion. Among them, the compression efficiency drops by 1∼2% as a whole compared with the dry condition, the operating point moves forward and the surge margin reduces slightly. The simulation results show that the presence of water droplets will change the distribution of airflow parameters along span direction, which leads to the decreases of mass flow rate and flow capacity of blade row, the increase of attack angel at tip region, and the separation of boundary layer in suction surface. Furthermore, the momentum losses caused by droplet impingement and breakup cause a sharp increase in the static entropy at the blade tip region as well. On the other hand, the ingestion of droplet brings an external disturbance to the flow, although it has some dissipated effects on the turbulence kinetic energy, it aggravates the unsteady characteristic of turbulent flow seriously at the tip region of blade.

Sweeping Jet Actuator (SJA) was introduced as a potential active flow control method for reducing three-dimension (3D) flow separations in a compressor cascade. Unlike some other actuators, SJA needs no valves or moving parts to convert its steady compressed air source into sweeping jets that oscillates from side to side through the millimeter-sized outlet nozzle. The rather simple and small structure makes it possible to place SJA into the blades. In this study, a 3D numerical simulation using unsteady RANS codes was conducted to investigate the effects of SJA on the flow pattern and the aerodynamic loss mechanism in a compressor cascade. Firstly, the reliability of a commercial Computational Fluid Dynamic (CFD) code was validated and the computed results showed good agreements with experimental data from the literature. Secondly, some possible affecting factors, such as actuating pressure, position of SJA exit and jet skew angle, were analyzed and discussed in detail. Moreover, the effectiveness of active flow control under different locations and stream directions of SJA was studied for obtaining a further understanding of the mechanism of SJA for controlling flow separations. In addition, the generation and interaction of internal secondary flows in the compressor cascade were also investigated, and the oscillating jet process of SJA was presented. The numerical results indicate that using SJA delays effectively the corner flow separation, thus decreases the aerodynamic loss of the compressor cascade. For the optimum scheme within the present research, the reduction of overall time-averaged total pressure loss coefficient achieves about 5.6% compared with the original case without SJA. The streamwise position of SJA has a more remarkable influence in improving performance than the other SJA schemes. The considerable improvements of flow separation in the corner region is considered to be one of the main reasons in overall performance increase.

Spline couplings which have simple structure, high reliability and can compensate torque transmission error are widely used in rotating machineries, such as aeroengine and gasturbine, etc. Recent efforts show that it is potential to make the rotor system losing its stability. Nevertheless, the experimental study of rotor system with spline coupling is rare and inadequate. This indicates a need to study the factors that affect the stability of rotor system with spline coupling experimentally. In this paper, a specially designed spline connection rotor test rig has been built and used to simulate a multi rotor system of turboshaft engine. The experimental instability characteristics of spline connected rotor system are presented. The instability speed and critical speed under different conditions such as lubrication conditions, external damping, load torque, spline tooth error and fit type of internal and external spline are measured. Based on the above-mentioned results, the effect rules of the influence factors on spline connected rotor system stability are studied. Results show that lubrication can effectively weaken the vibration of the system. The increased external damping makes the stability better when the spline coupling is unlubricated. With the increasing of load, the subharmonic vibration decreases after the system loses its stability, the system stability becomes better. The stability of spline coupled system with larger tooth error is better than that with normal one. Normal fit-up spline coupling improves the system stability under the conditions of lubrication and small external damping. This study may be helpful to get the favorable parameter setting of spline connected rotor system for avoiding instability and reducing vibration.