Service Communautaire d'Information sur la Recherche et le Développement - CORDIS


Executive Summary:
The AEGART project has developed and demonstrated a novel, beyond state-of-the-art type of starter-generator system for more-electric business jet.
The most popular state-of-the-art starter/generator technology employs a three-stage wound field synchronous generator with rotating rectifier. This machine has been extensively adopted in fixed wing and rotor craft applications and has proved to be highly reliable, inherently safe with voltage control achieved by varying the excitation current. However, the wound rotor technology with rotating diodes limits the machine speed hence the achievable power density, passive rectification to achieve dc output requires large and heavy filters and extra winding is required to perform starter function. Hence, the design cannot be regarded as optimal and have an obvious limitation if consider move to new, more-electric aircraft platforms.
Recent advances in areas of power electronics, electric machines and control methods made it possible to consider new starter-generator system topologies aiming for the improved performance, reduced fuel consumption and environmental impact, improved safety and reliability, as well as increased aircraft availability by implementing enhanced prognostic, diagnostic and health monitoring. The AEGART project has developed a starter-generator system proposing a radical, beyond state-of-the-art solution based on permanent magnet machine controlled by active front-end rectifier. Combination of unique properties of permanent magnet machine such as high power/torque density with fully controlled bi-directional power electronic converter in the main path of energy flow, thorough addressing the thermal management of both the machine and the converter, and development of advanced control and health monitoring strategies has led to the substantial improvement of overall system mass (about 20% reduction) and in system efficiency (about 10% improvement) compare to most advanced existing solutions. In addition, the AEGART system introduces new functionalities not employed in existing starter-generators, including:
- Controlled acceleration start making the starting duration independent on environmental conditions
- Capability of paralleled operation with other sources in future aircraft electric power systems architectures controlling the share of total load power taken by the system
- Capability to regenerate the excess of power in the DC grid
- Limitation of the torque slew rate seen by the engine
- Advanced machine fault diagnostic and fault prediction mechanisms to improve availability of the aircraft and to reduce unscheduled maintenance.

The developed AEGART system has been successfully tested in both University of Nottingham laboratories and at Labinal Power Systems’ ETB where all the key performance requirements/indicators set at the beginning of the project have been clearly demonstrated in the context of evaluation at TRL5.
Success of AEGART project means the development of a novel aircraft starter-generator system that allows saving weight (up to 20% compare to existing state-of-the-art systems) and reducing the environmental impact of next-generation aircrafts (by improving efficiency of electromechanical power conversion up to 10%), as well as improving their safety, reliability and reducing maintenance costs.

Project Context and Objectives:
The main objectives of the AEGART project have been set in Annex I as follows:
- Reduce the total weight of the starter/generator system by introducing novel overall system topology and new materials allowing for substantial increase of power density within the system components, and by optimized design for machine, converter and cooling system
- Increase the system safety by implementing fault tolerant starter/generator system design and developing health monitoring techniques to predict possible failures within the system
- Mature the technology by detailed design and manufacturing of the proposed system in the context of evaluation at TRL5 on the EDS electrical test rig.

The AEGART project has run from October 2011 till December 2015 (51 months). The key development phases included initial studies, design, manufacturing and testing. A summary description of the project context is given below:

- At the initial project phase, the main objective was to overview the potential topology candidates and to select the most promising machine-converter system for further detailed development. During this project phase, detailed requirements and specifications to the starter-generation system were studied by the UNOTT team, including limitations of current technologies. This was completed through intensive interactions with Dassault Aviation team. Based on the set of identified requirements and specifications, a number of possible starter-generator system topologies has been identified, their technical details were analysed and compared, and a short-listed topologies for subsequent trade-off studies were identified. This activity was followed by detailed trade-off studies, as a result of these the topology based on permanent-magnet machine controlled by three-level pulse-width modulated active power electronic converter was identified as the candidate for further development. At this stage the decision was to proceed to preliminary design with 2 machine variations depending on a rotor magnet system configuration (surface-mounted and interior-placed magnets). The final decision of Preliminary Design Review was to select the surface-mounted machine type for detailed development. However in parallel, as back-up solutions, the topologies based on induction machine and interior-placed magnets were further considered as well.

- Following the initial choice phase, the project entered into the design stage with the core objective to provide a detailed design of the selected topology candidate to meet all the requirements and specifications and potentially even go beyond them. Preliminary design of the selected topologies included detailed electric machine design, power electronic converter design, and design of thermal management systems for both machine (oil cooling) and converter (air cooling). These are reported in details in the project Deliverables. The initial designs were thoroughly discussed during the Preliminary Design Review (December, 2012). At this stage, a system topology for further, critical design development was identified – this is the topology based on surface-mounted permanent-magnet machine fed by 3-level IGBT converter. The design phase has been supported by significant modelling activities linked to electric machine design, power converter design, and thermal management system design. Significant attention at this project phase was given to overall system control design. The developed control concept is based on independent control of machine active and reactive powers through control in terms of synchronous rotating frame fixed to the rotor. In starting mode the advanced control strategy to minimise the machine current has been employed. In generating mode, the flux weakening to limit the machine output voltage at high-speed operation was implemented as well. The control design has delivered new knowledge of permanent-magnet machine operation in flux-weakening that can be applied to wide range of industrial applications. The project has also delivered an overall system model. The system compliance with the Requirements and Specifications was demonstrated at Critical Design Review. During the design phase, an innovative approach for management of machine faults has been proposed such that the machine turn-to-turn fault can be predicted by monitoring the insulation degradation stage. The corresponding ideas are patents-pending and enhance exploitation of the project results. This health-monitoring functionality requires application of specific wire configuration: this has been decided together with Dassault Aviation and the required wire has been manufactured. The project design phase has finished with the Critical Design Review (May, 2013) during which the detailed AEGART system design was approved and the system interfaces were fixed. Hence, the project could enter into the manufacturing stage.

- The main objective of the manufacturing phase was to provide a detailed mechanical design for the electric machine and its components, as well as for the power electronic converter, including its housing, associated harness and the control platform. According to the decision of PDR meeting, initially only the 1st system was built to ensure there are no errors and/or blunders in the design and to check the achievable system performance during the tests. Hence, manufacturing of the systems #2 and #3 was postponed until after the core tests of system #1 completed. The manufacturing of system #1 has been completed by July, 2014. The machine parts were manufactured mainly in-house, apart from the rotors and the casings – these were outsourced. The power electronic converter was build in-house. During the manufacturing phase, several visits to ETB have been undertaken by the Nottingham team to discuss and agree mechanical, oil/lubrication, electrical and control integration of AEGART system into the existing ETB configuration. One of the key elements at this project phase dealt with the development of the Acceptance Tests Program – this has been done in cooperation with Dassault Aviation and reported as one of the project Deliverables. The manufacturing phase has also included design and construction of the machine oil/lubrication system for subsequent tests in University of Nottingham laboratory.

- After the AEGART system #1 was manufactured, the project entered into the testing and validation phase with the key objectives to proof the achievable system performance, to validate the performance against the design requirements and specifications, and to demonstrate the AEGART concept in test facility environment. Integration of the AEGART system #1, including the developed machine, converter, control and oil/lubrication systems has been completed in September, 2014. Due to an extra care with the system #1 minimizing any potential risks and damages, safety precautions, and in order to gain the experience with the practical aspects of the developed system, the agreed tests program was progressing by small steps and the tests of the system #1 in Nottingham have been successfully completed in March, 2015. This approach to testing was also justified by the limited capabilities of ETB in terms of achieved speed range: as a result, the decision has been made to limit the system #1 speed to 20,000rpm in order to de-risk the system #1: the focus was given to key performance indicators like power range and efficiency, dc-bus control and dynamics, new functionalities etc. These key performance requirements have been confirmed experimentally and advantages over existing state-of-the-art system has been demonstrated. Following that, the AEGART system #1 has been delivered to ETB at Labinal Power Systems, Colombes (France). The tests at ETB repeated some key tests conducted in Nottingham in different "electrical" environment of the official "Copper Bird", and also included some advanced tests including engine starting from battery and parallel operation with other sources. The developed system has successfully confirmed all the key performances required. Using the feedback from the test campaign, AEGART team has verified, validated and optimised a set of simulation models of the system components and of the overall AEGART system developed at the design phase of the project. Since the tests campaigns were successful, the AEGART system #2 was manufactured as well: it will be employed in the follow-on programs within the frame of Clean Sky 2. The system #2 will also go through advanced test planned by full-range ATP, including maximum power capability, smooth transition into 540V dc operation, torque slew limitation and others. The key parts for AEGART #3 were manufactured as well and these will be kept as spares to support future studies.

Based on the above, the conclusion can be made that the AEGART project has successfully achieved all the key objectives. It means that a novel, beyond state-of-the-art type of aircraft electric starter-generation system is created, validated and demonstrated. The designed system type enables safe and reliable functioning and saves the overall aircraft weight, thus improving the environmental impact of aircraft operations. The project has also received wide dissemination, including journal and conference papers and presentations at the lead international conferences, as specified in the corresponding report Sections.

Project Results:

Due to the global efforts towards environmentally responcible air transportation, many aircraft systems technologies are currently undergoing significant changes. The state-of-the art technologies in the field are expected to be more fuel efficient, very safe, simpler in servicing and of easier maintainance. The way towards this goal has been identified as a move towards “more electric” aircraft by replacement of hydraulic and pneumatic sources of power with electrical counterparts. This can lead to an increased reliance on electrical power for a range of primary functions including actuation, deicing, cabin air-conditioning, and engine start. Hence an electric power generation system plays a key role in more-electric aircraft.

The current starter/generator (S/G) technology employs a three-stage wound field synchronous generator. This machine has been extensively adopted in fixed wing and rotor craft applications and has proved to be highly reliable, inherently safe with voltage control achieved by varying the excitation current. However, the wound rotor technology with rotating diodes limits the machine speed hence the achievable power density is limited, passive rectification to achieve dc output requires large and heavy filters in order to comply with power quality standards, and extra winding is typically required to perform starter function. Hence, the design cannot be regarded as optimal and have an obvious limitations if consider move to new, more-electric aircraft platforms. Recent advances in areas of power electronics, electric machines and control methods made it possible to consider new starter-generator system topologies aiming for the improved performance, reduced fuel consumption and environmental impact, improved safety and reliability, as well as increased aircraft availability by implementing enhanced prognostic, diagnostic and health monitoring. The AEGART project (stands for Aircraft Electrical Starter-Generation System with Active Rectification Technology) is targeting the development of such a novel system for business-jet platform. The project is a part of “Eco-Design for Systems” ITD within CleanSky JTI.

Consideration of many design aspects is required for this complex and multi-disciplinary task including definition of the best system topology, electrical machine and power electronic converter design, thermal management and cooling arrangements, development of control techniques, health monitoring and fault management systems. This part of the Final Report provides overview of the main results and achievements of the AEGART project and underlines the achievements beyond state-of-the-art.


The key technical requirements and specifications for the AEGART starter-generator (SG) system design are as follows:
In generating mode, the system should deliver 45kW of electrical power at the dc side (270V output)
Alternator maximum speed in generation mode – 32000rpm
Speed ratio in generating mode: 1.66, i.e. the lower generating speed is 19200rpm
In a generating mode, the system sets the output DC current as a function of network voltage according to the characteristics shown in Figure 1 (see the attached document where all the report illustrations are given). The required characteristic should be formulated by the following parameters: Vo – no-load voltage (rated value - 270V), required power Preq, maximum power Pmax and voltage drop ΔV.
The system provides engine acceleration for up to a self-sustained speed (rated value – 10,000rpm, maximum – 13,000rpm)
The engine starting can be provided at different environment conditions in accordance with the engine torque-speed characteristics for the ambient temperatures ranging from -40oC to +55oC as shown in Figure 2.
The starting is provided with a constant acceleration such that the acceleration time can be set within the range of 25…40s. At harsh environmental conditions (for example, extreme temperatures) the starting time may be increased. The system should also be able to provide full-torque start
The system provides an unsuccessful starting sequence (minimum 2 attempts) with 3 minutes pause between the attempts. A longer cooling-off period may be required after a second starting attempt under extreme environment conditions
The AEGART system allows for 5% overspeed margin
The design provides protection functions as detailed in the Topic Sheet

The Sections below reports how the AEGART team has developed the novel SG system to meet these requirements


In this Section the core idea of the project is reviewed. Consider a simple RL-electric circuit as shown in Figure 3(a). The circuit is connected to the sources VA and VB. If it is assumed that VA is fixed, and vary the magnitude and phase of the source VB, the current in the circuit can be controlled in both the magnitude and the phase with respect to VA. It means that the power flow through the circuit (active P and reactive Q) can be controlled in either direction. If the role of VA is given to the electric machine, and the role of VB – to the power electronic converter that is acting as a voltage source, then one can control the machine as a motor (directing the active power flow to machine) or as a generator (power is extracting from the machine) as illustrated by Figure 3(b).

The developed AEGART S/G system implements the described idea. Hence the developed SG system will consider some AC electric machine controlled by voltage-source power electronic converter. The topology of the machine-converter system was a subject of project trade-off studies.


The initial project stage dealt with the selection of the best possible system topology, including machine and converter type. Different machine types and converter topologies for S/G drive concepts and their associated thermal management, control and sensing requirements were considered. The capabilities of the different machines in achieving the desired characteristics and their respective properties formed the basis in determining the most suitable type and topology and verify the feasibility of the S/G. The machine types and topologies were assessed based on the required load and speed duty cycles and a comparison matrix was created in an effort to identify the most promising solutions. The main aspects of evaluation were as follows:
- Machine size, profile and weight
- Converter complexity and reliability
- Sensing requirements
- Peak torque
- Torque Ripple
- Extended Speed Range
- Reliability and Safety
- Adaption to cooling
- Scaling
- Technological Maturity
- Material cost and availability

As a result of initial considerations, the following topologies were selected for further detailed trade-off study:
Based on Induction Machine
Based on Switched Reluctance Machine
Based on Permanent-Magnet Machine (with surface-mounted magnets and with interior magnets)
In order to perform the trade-off study, each of the short-listed topology candidates has been subjectted to the process illustrated by Figure 4.

Initially, the identified machine types have been subject to a basic design analysis to establish the optimal aspect ratio, cross section, to guarantee a mechanical integrity, to define a number of turns to suit the required DC-side voltage, etc. Based on these criteria, the possibility of achieving a peak torque has been assessed. If failed – than the next approximation cycle has been initiated with modified initial selections. If the peak torque can be achieved – then the next stage of design has been undertaken: this includes calculation of machine losses and calculation of machine lumped parameters. In parallel, each machine type requires an appropriate converter topology and thermal arrangement – these have been included into the trade-off process as well, as Figure 4 shows. Machine lumped parameter model, together with the converter topology candidate, allows for calculation of converter losses and design of basic heatsink. Machine losses, together with possible machine cooling arrangements, allows for assessment of machine temperature – if this appears to be higher than allowed, then another cycle of basic design is required. If passed, then the total system mass and system efficiency can be evaluated, as well as system reliability and safety.

The summary of trade-off studies is given by Figure 5.
The benefits of system topology based on induction motor include robust machine construction, possibility to detect and manage the machine faults. The system will have good kVA/kW ratio and seems to be a good compromise between the performance and reliability. The major drawbacks include significant rotor losses (hence more load for thermal management system) and limited fault tolerance capability.
The topology based on SRM will lead to simple rotor geometry, the machine itself will be inherently safe and reliable, with good fault tolerance capability and inherent overcurrent protection, and low losses at high generation speed. From the other hand, the mass will be higher, and the kVA/kW ratio is high as well. In addition, this type of machine is likely to provide significant torque pulsations and produce high losses as the speed reduces (very high losses during starting mode).
The system based on surface-mounted PMM with metallic sleeve will have low mass and high performance, and can be designed for fault tolerance. However there will be significant rotor losses (in sleeve and in magnets). Faults are not easily detected in this system type, and the kVA/kW ration is high. If the sleeve is made of carbon-fibre material, the rotor losses will be reduced, however the design will be less robust mechanically. The surface-mounted machine may be designed for operation in flux weakening, but in this case the losses in the machine will be higher due to increased stator current (additional defluxing component). However the losses on Power Electronic Converter can be noticeably reduced.
The main benefits of topology based on interior PM machine include low mass and low kVA/kW ratio, as well as possibility to increase machine torque due to exploiting saliency torque component and possibility to implement flux-weakening control. However this structure will have a limited fault-tolerance capability and detection of machine winding faults will require a special attention.
All the topologies were considered with 2-level and 3-level Power Electronic Converter: this is indicated as column 2L and 3L in the Figure 5.

Following intensive discussion over the possible solutions at the Trade-off Study Conclusion Meeting, the selected topology candidate for the future development was a PMM-based one with IM-based topology as a back-up solution. It was decided that for Preliminary Design the AEGART research team will prepare 2 solutions:
- SM PMM with flux weakening and 3L converter (column C32 in Figure 5) and
- IPM with 3L converter (column C42).
These two solutions are very close by design, hence in a case of unpredicted design or manufacturing problems/issues there would be an easy way to switch to another solution thus reducing a risk for a smooth project flow. As a final back-up, the IM-based topology was designed in more details until the PDR stage.


This section overviews the design of the AEGART electric machine as the core element of the SG system. Figure 6 illustrates the required torque-speed characteristic. The machine runs as a motor during engine start and must supply constant torque from standstill to upper motoring speed (around 8,000 rpm). When the machine reaches the lower generation speed ωmin, the system turns into a generating mode and should supply the required power to the loads.
An SPM candidate was chosen considering mechanical and thermal constrains at high speed, power density, reliability and size. In the design process, fault tolerance is considered by adopting a redundancy solution in the event of a failure. This allows a distributed winding to be adopted, consequently minimizing rotor losses in both magnets and in the rotor back-iron. Different slot-pole combinations were investigated in view of the overall losses and performance and the compromise in the design of the machine for operating as an engine starter and as generator. Implications of different magnet’s retention material and magnetic materials were studied as well as overviewed below.

Slot/pole combinations

The slot (Q) /pole (p) combinations considered during design are shown in Figure 7. Both single and double layer (Nlay) winding topologies are considered in the selection procedure. The number of rotor poles is limited to 4, 6 and 8 as any higher number of poles will lead to high electrical frequencies and higher power converter losses. In order to limit the temperature increase in the rotor as a consequence of the eddy currents induced by MMF harmonics at high speed, the winding topology for each slot/pole combination is selected such that the winding harmonics are minimized for an acceptable trade-off with the fundamental winding factor.
Figure 8 represents the winding factors associated with different slot-per-pole-per-phase (q) and winding layer number combinations considered in the comparison designs, where the parameter ᴦc represents the coil pitch.
In the design, the stator tooth maximum flux density at no-load is considered to have 75% of the saturation flux density of the core material. The total weight of each machine is the same. This includes the stator, rotor and the copper associated with the main winding and the end windings. Each machine is thus designed for a constant weight and fixed outer diameter. The selection is made considering the efficiency and performance of the design while it satisfies mechanical and thermal constraints as explained below.

Magnet retention mechanism

In order to safely retain the magnet at high speed operation two different sleeve materials are investigated: Carbon Fibre (CF) and Inconel 718. At the preliminary design stage, the sleeve thickness calculation is performed based on a 150% over-speed safety margin. This margin of safety has been revised to 110% and the sleeve thickness requirements have been re-evaluated. Evaluated sleeve thickness requirement for the selected rotor radii and magnet thicknesses are presented in Figure 9.
From Figure 9, it is seen that the required sleeve thickness is highly influenced by rotor radius as centrifugal force proportionally increases with the rotor radius; however the influence of the magnet thickness over the sleeve requirement is insignificant. Table in Figure 10 shows the pre-stress of the 4 mm sleeve due to application of CF and Inconel at maximum speed. Both provide reasonable prestress with corresponding sleeve undersize and provide the capability to design a machine with the required split ratio of 0.54. However CF has been chosen for the design since it has negligible eddy current losses compared to Inconel sleeve as shown in that table.

Magnetic material selection

Figure 11a illustrates the B-H characteristics of 2 materials considered: JNHF silicon steel and Cobalt alloy (Vacoflux-50) and Figures 11b and 11c show the corresponding flux density plot, respectively. These two materials are employed in multiple machine design and analysed. Different designs were considered and combined into a Table that is too awkward to be placed in this paper, however will be presented during the conference. As discovered, the highest efficiency can be achieved by machines with higher pole numbers. However, with consideration on the power electronic losses and high frequency control limitations, the 6-pole machine designs was considered for the final machine design.
Out of the 6-pole machine variants, the best machine candidate was selected as 6-poles 36-slots as it has the lowest eddy-current losses. The machine design with more expensive cobalt based alloy Vacoflux-50 does not achieve lower losses in comparison with the cheaper JNHF design during the continuous duty generating points ωmin and ωmax. Hence the JNHF material is adopted for this starter-generator application.

Thermal management

Direct oil cooling of the machine has been identified as the optimal solution. The core idea of machine cooling is to use ducts through the stator core along the stator outer diameter and slots existing between stator teeth, as illustrated by Figure 12. Such arrangements guarantee an even distribution of the coolant. Further enhancing cooling techniques have been identified for the stator and rotor regions aimed to effectively reduce the machine weight and improve system efficiency. Numerical simulations using a CFD code were conducted to investigate the fluid flow inside the machine. Another numerical method, based on the lumped parameters thermal network has allowed getting detailed information about the temperature distribution within the machine.

Machine mechanical design

Following the electromagnetic and structural design, the electric machine mechanical design has been performed. The details of the design were reported in corresponding Deliverable, and the exploded machine view is depicted in Figure 13. When built, the machine measured mass was 20.1kg. The analysis of mass constituencies is illustrated by Figure 14. At the Figure 15 one can see few photos of the manufactured machine.


From the machine preliminary design, the boundaries for the converter were defined as 400A peak output current, 1.2kV DC-link peak and an electrical frequency equal to 1.6 kHz, given the three pole pairs for the machine. In order to design the power converter, the chosen candidates were the two-level and three-level NPC inverter. The two-level was considered for its known simplicity and intrinsic reliability due to the minimal number of devices. The three level was also considered because it brings advantages in terms of lower EMI emissions and it will also allow higher fundamental frequencies for the same switching frequency. This is a useful feature when considering the high speed application of this drive. The final decision, however, come from the power losses comparison for the two topologies, simulated for a similar load condition and THD. Minimizing the losses is important in aerospace because it will lead to a minimal heat-sink that has a significant influence on the overall mass. Such simulation was performed importing the thermal characteristics of actual suitable devices and calculating the total losses for different speed and load conditions. Solutions with MOSFET were also considered, but even providing an excellent performance in terms of losses, they have been discarded.

Comparative loss analysis

The first losses simulations were performed comparing a two level inverter based on a two level Infineon power module rated at 400A and 1.2kV, and a three level one, rated at 400A 650V. Figure 8 shows the comparison of losses for these two converters, operating respectively at 20kHz and 16kHz. It was found that these were the two switching frequency needed to have a similar THD for the same load in the two converters. Figure 16 compares three operating speeds, the lowest being the top speed for motoring mode, while the other two are the two extremes for the generating mode. From Figure 16 it is clear that for the speed ωstart in motoring mode, the both cases have similar values, however their components are quite different. The two level has the major component spread across conduction and switching of the IGBTs; while, for the two level that has devices for lower voltage, the switching in not so relevant. The three level however, having two devices in series, has a dominant conduction component. In generating mode the current will flow mostly through the diodes, and again this is clear from Figure 16, where the IGBTs conduction losses are much smaller for both the cases; for this mode of operation the losses are favourable in the three level topology, even though the detail of the conduction losses for the diodes are much bigger, again for having two devices in series.
One more interesting simulation was carried out to evaluate the losses considering MOSFET devices, Infineon CoolMOS™ IPW65R037C6, considering to parallel 10 or 20 of them in a three level system, as it can be seen in Figure 17. The MOS solution will provide great advantages in terms of performance; the number of devices, however, will be 120 or 240 respectively for the 2 options, without considering the six extra diodes. This would create a serious problem when designing the layout of the power modules unless a custom power module would be manufactured; hence this solution is considered as impractical. Another consideration against this solution was that the high number of devices would increase significantly the level of complexity for the system, impacting negatively on the reliability of the overall system.

Converter hardware design

For the hardware design the 3-level converter structure (Figure 18) was adopted and the chosen IGBTs were the Infineon F3L400R07ME4, rated at 400A, 650V. In order to provide a stiff DC-Link, but at the same time a small and light solution, a custom design has been adopted: a single enclosure for the twin capacitors rated at 600µF, 650V, 200Arms. The control board is based on the Texas Instruments DSK6713 running at 225MHz coupled with an Actel FPGA ProASIC3 A3P400. The gate drives boards use the Infineon driver 1ED020I12-F, they have been specifically designed to fit the layout of the power modules, minimizing the connections but at the same time providing single driver status and real time temperature monitoring for every power module.
A basic hardware dead time network is introduced to minimize the number of digital connections, as shown in Figure 19. All the digital signals within the control board are transmitted by just nine optic fibers: 6 to control the IGBTs plus 3 for status monitoring. The final layout of the converter can be seen in Figure 20.

Converter thermal management

The design of converter’ thermal management was based on calculated losses in the thermally worst case of repeated start scenario (unsuccessful start). As a result, an forced air-cooled solution was employed and the heatsink for power modules was designed to guarantee an acceptable temperature at the end of the worst-case scenario. Following the calculations, the system will be able to provide an infinite number of repeated start attempts.
Figure 21 shows the box designed for AEGART power converter cooling. In the centre of the box, there was a heatsink integrated with the air duct; the converter is attached to the baseplate of heatsink. Control electronics and a large capacitor were then attached to the power converter. At end of the air duct, there is a fan which sucked air through the fins. The air duct, capacitor and heatsink are fixed on a large baseplate and then covered by an aluminium enclosure. The front and rear plates are made of grid panels to enable some ventilation. Air was sucked through front panel, entering the heatsink and then discharged from the fan outlet. This design is not optimised for mass and intended for laboratory tests only. The total mass of the converter with box is around 27.8kg. The fan selected to drive the air is Rotron SPARTAN Tubeaxial Fans (part number of 011683000), the power for the fan is driven by 28V dc and power consumption is 33W.
For the demonstration purpose, an aircraft-fit converter version was designed as well and demonstrated by CAD drawings. This is illustrated by Figure 22: the box accommodates two converters sharing the same heatsink. The indicative power density of this solution is in the region of 4kW/kg, and the mass analysis is reported in Figure 22 as well.

Overall view of the manufactured AEGART converter is given by pictures in Figure 23.


A control structure to ensure smooth system operation through the entire speed range was designed. The proposed approach expands flux-weakening operation into a generating mode and allows for the system to act as a stand-alone or paralleled source in an on-board electric power system.
As it follows from the starter/generator system purpose, the control target depends on the operational mode. In motoring mode, the system should provide a mechanical torque to accelerate the engine up to the firing speed. When the engine runs, the system should control the output voltage (in this study – dc) to feed the onboard loads. Hence, the control structure is also different. This is considered below. However, both control structures will employ the same “core system” based on a SPM controlled in terms of the synchronous dq reference frame, aligned with the rotor, as Figure 24 depicts. Fundamental vector control is used in the core system to control machine d- and q- currents according to the demands.
To perform a motoring function, a standard SPM drive system structure, including a flux-weakening control algorithm, as shown in Figure 25, is employed. When the stator voltage is less than the maximum value, the output of the controller is saturated to zero. The controller activates when the stator voltage exceeds the maximum voltage limit. With the outer voltage regulation loop, the flux-weakening controller adjusts the current angle γ to inject the negative d-axis current thus de-fluxing the machine at high speeds.
In generating mode, the machine speed is no longer a controlled value but is an external parameter to the system which is determined by the engine. The task in this mode is to control the dc-link parameter, typically the dc-link voltage Vdc. The reported system (given in Figure 26) is intended for operation in parallel to the other sources; hence the Idc control is required according to the droop characteristic. In this mode the flux weakening control remains the same as for the motoring mode, however an Idc controller is employed which affects the machine iq demand. The Idc reference is generated by the droop calculator based on an actual Vdc measurement.
Rigorous control system design has been performed, together with detailed simulations, to confirm the system performance and stability throughout the operational range. As it was shown and reported through the published paper, an improper selection of the controller parameters may easily lead to an unstable system operation.
Initially the designed control was confirmed by time-domain simulations. For example, results if Figure 27 shows the scenario in which generator speed varies in a wide range whilst the dc-bus connected loads change from zero to full rated power. As one can see, as speed changes, the system provides good control of dc-link voltage and current according to the required droop characteristics. More simulation results to demonstrate the system performance, including in starting mode, were presented and discussed during the project dissemination activities (conferences, meetings, publications).


The key system performances were successfully demonstrated during the tests campaign, initially in University of Nottingham laboratory, followed by tests at ETB.
At the beginning of the tests, the decision has been made to go through reduced test campaign with the system #1. The main reason was due to limited capabilities of ETB facility in terms of achieved speed, so the tests were limited to 20,000rpm range. From the other hand, it allowed to de-risk the project, to gain a good experience and confidence in the system operation and focus on key performance indicators as well as on novel functionalities offered by AEGART, i.e. to confirm the development of beyond state-of-the-art solution. More advanced test, such as demonstration of achievable maximum power, operation in extended dc-voltage range, achievable dc-bus control dynamics, torque slew-rate limitation and some others were left for the test campaign of system #2 (within the follow-on Clean-Sky-2 project EMINEO, in which further enhancements and EPS solutions based on AEGART output are to be studied).
At the initial testing phase, the AEGART system has gone through intensive test campaign to confirm core achieved performances. In summary, all these were successfully met, and the full report on UNOTT tests was submitted as project deliverable. Here only the key achievements are overviewed.
The tests were conducted according to “Acceptance Test Plan”. Examples of some rest scripts results are given in the Figures 28-35 below. From the measured and calculated back-EMF for the electric machine at 8000rpm (Figure 28) one can see very good match of the experimental curve to the machine analytical design. Machines short-circuit current and breaking torque reported in Figure 29. As it follows, both of these demonstrate very good match to the expected from the design as well. Figure 30 shows waveforms of machine currents and converter voltage at 9000rpm/40Nm operation. As one can conclude, the machine current is very close to sinusoidal, and the converter clearly demonstrates 3-level operation.
In Figures 31 and 32 AEGART system performance in the starting mode is proved. It is clearly demonstrated, that the system provides constant acceleration starting and the engine start-up time is independent on environmental temperature: in both cold (-40oC) and hot (+55oC) conditions it takes 40s to accelerate the engine to 10,000rpm. The desirable time (25s…60s+) can be set according to the pilot preference or other requirements. This is one of the new functionalities provided by AEGART system which is not available in the existing systems.
Another new functionality offered by AEGART its capability to operate in future electric power systems architectures with parallel operation of multiple sources. This can be achieved by controlled output characteristics (droop-control). Figure 33 confirms that the designed system provides the characteristics as requested by the user. Note that system is also capable of energy regeneration and sending it back to the engine (in Figure 33 this is the area of negative Idc values).
The tests also included investigation into the system response to changes of the droop references. This will allow to the AEGART system to control the amount of taken load power. Some example responses are given in Figure 34. This tests shows the unique functionality of AEGART – capability to manage delivered electric power when operating in parallel with other sources: this is one of the key trend in state-of-the-art aircraft electric power systems development aiming for future platforms, and AEGART is ready to support it.
Another important issue investigated during the tests was a system torque ripple: this has also been analysed; the spectrum was assessed looking the torque-component of the stator current and the results are shown by Figure 35. The designed system provides very favourable level of torque pulsations at the engine shaft.
Following completion of tests program in Nottingham labs, the system was delivered to ETB for further investigations using Copper Bird® environment. The core of the test program was similar to the tests conducted in UNOTT but with the speed only up to 19,000rpm due to limitations of ETB test rig. From the other hand, the ETB the tests campaign has included confirmation of more advanced system capabilities including:
- Controlled acceleration starting using batteries power
- Parallel operation with other dc source (entering into parallelism and contribution into the overall power budget)
- Control of load power sharing with other source by manipulating the AEGART system droop characteristic.
As a summary on testing, AEGART system has successfully confirmed the design requirements, new functionalities and demonstrated the expected performance. The system is clearly demonstrated its beyond state-of-the-art status.


This Section of the report reviews the excellence of AEGART project. Summarising the above, in technical, technological and scientific achievements present a significant step beyond the state-of-the-art in starter-generator systems. The system developed by AEGART project has set new key figures for starter/generation systems:
The overall system weight is reduced by 16-18% compare to existing systems (based on Dassault Aviation expertise).
The overall system efficiency in electromechanical power conversion (from input shaft to output dc bus) is improved by 8-10% (based on Dassault Aviation expertise).
With these advantages, the AEGART is clearly contributing into development of sustainable, greener aviation.
AEGART project has also generated ideas for potential follow-on programs. Among these, the following should be noted:
Investigation into technologies of machine wiring implementations in order to demonstrate safety of permanent-magnet machines as a main on-board electric energy source. Within AEGART, the concept has been developed and partially demonstrated, however due to its importance more thorough and detailed investigations required
Application of new semiconductors into AEGART power electronic converter, such as SiC and GaN devices, in order to increase sharply the converter power density and cooling requirements. This will potentially lead to a very compact solution. Integration of the converter into machine body with common cooling circuits may also lead to significant weigh/volume benefits in future systems
Consideration of novel machine types, including hybrid machines topologies and machines with high-temperature windings (including new composite materials). This will also entail further advances in control and thermal management systems.

Based on above, the conclusion is that the AEGART was a very successful project with significant impacts: these are detailed in corresponding report Sections.

Potential Impact:

The AEGART project has significant impact that spans technical/technological, scientific, socio-economic and societal components.


The key technical/technological impact can be defined as design, development and demonstration of beyond state-of-the-art aircraft starter/generation system that:
- Removes limitations of current technologies towards more ecologic aircraft
- Reduces the total system weight and volume increasing efficiency and ensuring reliable, safe and low-maintenance operation.

The system developed by AEGART project has set new key figures for starter/generation systems reducing their overall weight by 16-18% compare to existing systems and improving their efficiency by 8-10%.

These key achievements have been reached by introduction and exploitation of an innovative technical concept based on actively controlled power electronics into the main path of electric energy flow:
- introduction of active converter allowed for an improved electric machine design carefully addressing machine efficiency and losses in both starting and generating modes, as well as machine torque density, thermal dissipation and cooling arrangements
- employment of active converter improves the efficiency of power conversion and output electric power quality compared to existing autotransformer-rectifier solutions thus reducing significantly the size, weight and volume of required passive filters, up to their full elimination.
- Introducing a switched power conversion in the main power path, the project addressed issues of fault tolerance, health monitoring, system integrity and availability. Thus, the project has developed understanding and established a set of qualitative criteria for in-path power conversion. This experience and knowledge will have significant benefit for future projects in the field.

The overall technical impact and excellence of AEGART project is built-up through the following constituencies:
- New paradigm for the starter/generator system to overcome limitations of current technology towards more efficient greener aircraft. It is shown that the starter/generator system based on permanent-magnet electrical machine is a viable solution for aerospace electric power generation that allows for substantial improvement of key achievable performances. The benefits of the proposed system topology are clearly shown in comparison to other potential topologies, including state-of-the-art topology based on 3-stage field-controlled synchronous machine, as well as induction and switched-reluctance machines
- The developed starter/generator presents a significant step not only in reducing the overall system mass/volume and increasing efficiency, but also introducing new functionalities never seen in previous systems. These include engine starting with controlled acceleration, capability to work either as a stand-alone source of power or to run in parallel with other sources, control of the load power share taken by the system, torque slew-rate limitation, machine winding health monitoring and fault prediction
- The project has demonstrated new technological solutions as for high-efficient and high-speed permanent-magnet machine design (active stator cooling by flooding it whilst the rotor area remains dry in order to build efficient thermal management system, employing carbon-fibre sleeve for rotor magnet retention), converter design (multilevel topology, modulation strategy, devices), as well as a novel control concept for a machine-converter system (high-speed operation in flux-weakening mode, droop control, controllers design)
- The project has developed a set of simulation models to represent key system components at different fidelity levels. The software was delivered to the Topic Manager as open source models and will be exploited in the current and future research programs
- The project has also provided support to the industrial testing and demonstrations of the developed system


In addition to the project technological/technical impact(s) mentioned above, there is also a significant scientific impact of AEGART. This can be summarised as follows:
- The project has enhanced understanding of design approaches and achievable performances of permanent-magnet electric machines. These include both electromagnetic design, structural design and thermal management, as well as better understanding of the ways of manufacturing such a machine when aiming for its ultimate performance
- The project has developed a unique method to monitor health of permanent machine winding - the known drawback of this machine type, namely a short-circuit issue limiting the application of this high-performing machine type in aerospace can from now be adequately addressed
- New knowledge for flux-weakening operation of permanent-magnet machine in generating mode has been developed. The project reports a rigorous control design to ensure high-performance, safe and reliable machine operation through the wide range of engine speeds
- The project has also contributed into the development of novel aircraft electric power system architecture paradigms and has developed principles of paralleled operation of multiple sources acting on a single bus, ways of establishing desirable power sharing ratio and stability conditions for such systems when supplying complex and non-linear loads

Summarising the above, in technical, technological and scientific directions the AEGART project outputs present a significant step beyond the state-of-the-art in such areas as aircraft electric machine design, power electronic conversion systems, their thermal management, as well as advanced control design and implementation.
The project results contribute towards the competitiveness of the European providers by achieving a significant weight and cost saving for future more-electric aircraft, whilst ensuring the electrical generation system is capable of reliable, effective, safe and low-maintenance operation.


The socio-economic impact of the project has been discussed with the Topic Manager, including ways of commercializing of the developed AEGART system. If moving the system into the business-jet market, many 100’s of units will be required, hence a significant number of jobs can be potentially created. In the project team’s experience, there are [x] clear manufacturing steps to produce such a motor, each requiring [x] number of skilled people. If this was implemented on a commercial line, it would result in at least [y] jobs created.
Furthermore, the same concept can be applied to other aircraft platforms requiring design modifications based on the overall architecture and concepts demonstrated in AEGART. This will generate at least 5 further highly skilled jobs covering mechanical design, electrical design, cooling system design, design for manufacture and overall integration of the systems – in this case the impact will be substantially increased.


In terms of the project societal impact, the developed new aircraft system will lead to more efficient and lightweight solution hence will benefit to the community by significant reduction of pollution due to reduced fuel burn of future aircraft. The 16-18% weight reduction compared to current technology represents a significant fuel saving and therefore corresponding savings in CO2 and NOx.
In addition, the project outcomes will lead to a system that improves safety and comfort of the flight hence well-being of the passengers.

DISSEMINATION of project results:

The project and its key findings were widely disseminated. There are 17 publications at the leading international Conferences for which the papers passed through a peer-review process. The full list of papers is given in the corresponding report field.
Currently 3 journal papers are being prepared for submission to IEEE Transactions on Aerospace and Electronic Systems, and for IEEE Transactions on Industrial Electronics.
At the time of this report writing, 3 patent applications linked to the project are being prepared.

The AEGART project was also represented at the Clean Sky Conference (Brussels, March 2014), “More-Electric Aircraft” 2nd and 3rd Conferences (Hamburg, 2014 and 2015, correspondingly). These conferences were attended by worldwide aerospace companies and the project has attracted significant interest.

Project has also an academic interest – the key project elements and findings are introduced into lectures in the University of Nottingham as part of the MEng and BEng in Electrical and Electronic Engineering; students are also given topics for their final year projects aligned with the AEGART scientific direction.

Project Principal Investigator, Dr S Bozhko, and WP4 leader, Prof P Wheeler are members of SAE AE-7 Standards Committee, and the project outcomes will be discussed at the following meetings and will effect work on future standardization documents on systems for next-generation aircrafts (within SAE AE-7C Committee)


In terms of exploitation of project results, the key will be maturization and commercialization of the developed system. Currently the machine is demonstrated at TRL5, and power electronic converter – at TRL4. Successful industrialization of the project outcomes will allow increasing the achieved TRL levels and will gradually result in a certified flying product. Nottingham University is keen to support Dassault Aviation efforts towards industrialization of the AEGART system.

A significant importance should be given to the lessons learnt during the project development. These are related to particularities of high-efficient machine design, power electronic conversion topology, control platform design, thermal managements system, and detailed mechanical design and manufacturing of the AEGART system. The acquired knowledge and experience will allow transferring of these lessons for other platforms, for example – to rotorcrafts or large passenger aircrafts, as well as to into other areas such as automotive applications, railways, marine and others.

List of Websites:
Main contact:
Dr Serhiy Bozhko
Associate Professor in Aircraft Electric Power Systems
Aerospace Technology Centre
Triumph Road, Jubilee Campus
University of Nottingham
Nottingham NG7 2TU
United Kingdom

Phone: +44 (0)115 8468490


Jill Harris
Tél.: +44 115 8466757
Numéro d'enregistrement: 183664 / Dernière mise à jour le: 2016-06-06
Source d'information: SESAM