The Growth Engine Concept and Its Potential for an Electrified Aviation Future

The design of a clean-sheet gas turbine engine constitutes a massive development program worth more than a billion dollars, and may take place only once in a decade or so [1]. Three factors can determine whether a clean-sheet engine design should be realized or not: time, cost, and performance [2]. If the goal is the best possible performance ignoring associated cost penalties, then clean-sheet engine design is preferred. If delivery time is critical and performance compromises can be accepted, then an up- or derated derivative engine can be a sufficient solution. However, the current situation in aviation pushes original equipment manufacturers (OEMs) to mitigate technical risk, improve engine performance, accelerate delivery time, while at the same time reducing development costs across all engine programs. Within this context, the concept of growth engines has proven successful. In the literature, it may also be referred as the concept of core commonality or as the generic concept of engine families.

This approach primarily focuses on the utilization of a geometrically fixed or geometrically similar engine core across a family of engines. The core of a gas turbine engine comprises the high pressure (HP) system, namely, high pressure compressor (HPC), combustion chamber, high pressure turbine (HPT), and the in-between ducts, as illustrated in Fig. 1 for a typical two-spool turbofan engine. For this specific case, the low pressure (LP) system consists of the fan, the booster, and the low pressure turbine (LPT). For other engine architectures, there might be an intermediate pressure (IP) spool, connecting the intermediate pressure compressor (IPC) with the intermediate pressure turbine.

The definition of the growth engine concept holds true in either of the following cases:

1. The core used among engine variants of the same family has a fixed geometry with possible changes in thrust ratings; defined as geometrically fixed core,

2. The core used among engine variants of the same family has a modified core design with possible changes in geometry, core power and temperature levels that follow certain design rules to ensure commonality; defined as common or geometrically similar core.

Core commonality (or similarity) is ensured if the core compressor exit corrected mass flow is kept constant across a family of engines [3]. Another definition followed by industry and proposed by Kurzke et al. (1) is the term of core compressor mass flow at inlet conditions, corrected by pressure and temperature at exit conditions. Their only difference is based on choosing either the inlet or the exit absolute mass flow, thus ignoring or considering potential core compressor bleeds. Keeping either of the two terms at contant levels results in fixed compressor exit area and certain blade height of the last compressor stage. Furthermore, the design point of the HPC, defined by corrected speed and beta-lines, should be the same to ensure core commonality between the baseline core and the upgraded common core engine variants.

In more detail, the core commonality rules have been described by Sands [4] with four numerical practices that should be incorporated within the engine performance tool:

1. Vary the desired HPC inlet mass flow scalar to achieve the new compressor inlet corrected mass flow level (if scale of core in-flow equals unity, the new capacity equals the baseline one),

2. Vary the HPC design pressure ratio scalar to keep the compressor exit corrected mass flow at constant levels (an increase in core in-flow while keeping the compressor exit corrected flow constant would require higher HPC design pressure ratio),

3. Vary the HPT flow capacity scalar to maintain the same HPC beta-line value at design point (along with setting the desired design corrected speed yields fixed HPC operation at design point), and

4. Vary the HPC efficiency scalar to accommodate the change in polytropic efficiency in case of HPC technology infusion.

For the case of a geometrically fixed core (case 1), design changes in the LP and/or IP system (if present) are allowed to achieve different thrust ratings. These changes can range from adding or removing stages to even completely redesigning LP and/or IP components [4–8]. “Supercharging” the HP spool without changing the HP components geometry can also be achieved by modifying the booster (or IPC for a three-shaft configuration). Increasing the booster pressure ratio and mass flow at constant levels of HPC flow capacity result in higher LPT mass flow at fixed flow capacity. This yields more core power and hence, increased thrust ratings.

On the other hand, upgraded or geometrically similar core design (case 2) can be achieved either by changing operating temperatures of the core (T3, T4) or by varying core in-flow [4]. The latter can be realized through zero-staging the HPC (adding a new compressor stage before the original first stage) and flaring the HPC. Flaring without zero-staging can be an option only if the HPC was sized to operate at less demanding conditions than maximum loading. Adding one compressor stage can also be the case if the stage loading of the new variant surpasses the acceptable limits. Of course, design changes in the LP and/or IP systems are also possible for the common core case (case 2), as described above.

Technology infusion can be considered as the main driver of fixed or common core designs for multiple applications introduced over time [4]. Such technology upgrade packages can be realized in terms of turbomachinery aerodynamics, and material/manufacturing improvements [5,6]. The former can be represented by polytropic efficiency deltas at design point, added to the baseline efficiency values. Component weight reduction can reflect on the effect of material or manufacturing improvements. Technology upgrade packages can be applied to the LP/IP systems in a geometrically fixed core concept (case 1). On the other hand, the common core concept (case 2) allows technology infusion to both, the LP/IP systems and the engine core.

It should be clear that, in most cases, the growth engine concept imposes penalties on cycle performance of the variant engine compared to an individually optimized core of a clean-sheet engine. It is reported that a small technical risk is always associated for this reason [9]. A common engine core is designed to serve different engine cycles encompassing different application-specific limitations and constraints. Engine manufacturers are called to mitigate technical risks as much as possible, while ensuring that the provided engine variant offers a maximum return on the investment of the initial development program. It was proposed by Sands [4] that the confidence in their design decisions for the initial development of a new engine program must be proportional with the capital risk involved.

The major advantage of the common core concept lies on the fact that the proposed engine variants will be used among different aircraft types, distributing the massive investment cost across several programs [5]. Lehmann [3] reported that if an engine core is sufficiently common across different engine applications, its development can start early, and even prior to the selection of the first aircraft system to be installed. This substantially reduces the development time of each engine variant. Considerable drop in development, production, and acquisition costs have been shown [3] when a large amount of common core parts and product lines are shared among the engine variants. Furthermore, reduced cost of spare parts, standardization of maintenance procedures, enhanced flexibility, and responsiveness of manufacturing processes lead to ownership cost savings.

Even though the concept of growth engines is not new in the aero-engine industry, it has not been investigated systematically in the open literature. This provides an excellent opportunity to shed light into the foundation of this concept, showcasing the engineering theory that is based upon, along with exemplary analysis of realistic conventional engine applications. The research interest for an electrified aviation future provides high potential to implement the growth engine concept to the design of hybrid-electric propulsion architectures. However, the imposed design challenges by electrification itself should be scrutinized in order to realize a family of electrified propulsion systems.

The remaining article is structured as follows: Sec. 2 introduces two of the most successful aero-engine programs in aviation history, and explains the development plan of each manufacturer that implemented the growth engine concept. Section 3 provides insight into the origins of the growth engine concept and the theoretical definitions from a generic engineering perspective. Section 4 analyses the methods and associated tools to initiate, design, and evaluate an engine family program. In Sec. 5, the challenges and opportunities due to electrification are discussed, providing technical insight on a potential application of electrified family of engines. Finally, Sec. 6 outlines the major findings of this review work and proposes future research directions to exploit this proven concept for innovative aircraft propulsion applications incorporating electrification.

There is no doubt that the CFM56 program represents the largest and most successful international commercial aircraft engine program in the current world. It constitutes an excellent product family example, where strategic decisions and thoroughly scrutinized actions were taken, even before the first engine variant was released. A brief historic retrospect of the CFM56 engine family will follow, highlighting key technical decisions on each engine variant application and justifying them with publicly available data.

During the early 1970s, Societe Nationale d'Etude et de-Construction de-Moteurs d'Aviation (SNECMA) directed a strategic synergistic move toward the development of modern, fuel-efficient, and low-emission commercial engines in the “10 ton” thrust class [10]. These engines would compete against the already established JT3D and JT8D turbofan engines. Three major engine manufacturers (Rolls Royce, Pratt & Whitney, and General Electric) were approached to get the 50% partnership with SNECMA. General Electric (GE) was selected for the aforementioned engine program, and a joint company, called CFM International (CFMI) was formed in 1974. GE had already developed the F101 engine core, while SNECMA had a preliminary design of the M56 turbofan engine generating 20,000 lbf or 89 kN of thrust [11]. As far as the program organization is concerned, GE was responsible for the engine core, the main engine control and the system design integration, while SNECMA was in charge of the fan, LPT and accessories design, as well as the engine installation [10]. More information on the joint company organization and work split can be found in Refs. [12–14].

Retaining the same engine architecture and design philosophy across all family products were a decision taken from the very beginning of this program. In more detail, a two-spool direct-drive turbofan engine comprising a single stage fan, an low pressure compressor (LPC) (or booster), an engine core, and an LPT—driving both the LPC and the fan—was fixed for each engine variant. The engine exhaust could have either separate or mixed flow ducts, depending on the application. CFMI knew that in order to enter the single aisle aircraft market they had to provide an engine with exceptionally good performance and high reliability, while also considering engine growth potential for future market demands. Their design philosophy was based on technology advancements through evolution rather than revolution [12]. Specifically, Fig. 2 illustrates the CFM56 development plan in a qualitative figure of core technology levels as a function of entry into service (EIS), highlighting the key features of each engine variant.

During the mid 1970s, GE had already developed a powerful and compact core for the F101 engine, comprising a highly loaded compressor and turbine, as well as a short and low-emission combustor, leading to the highest thrust-to-weight ratio in its class [12]. The first prototype product of CFMI was the CFM56-2 engine which was tested in 1974, certified in 1979, and entered the market in 1982 re-engining the DC8-70 series. This variant had exactly the same core as the F101 engine, exploiting GE's advanced core thermodynamic cycle. The key feature of the CFM56 program strategy was the testing of each engine variant at the most severe conditions and entering into service at derated thrust levels. Core testing at higher than planned-to-operate ratings, temperatures, speeds, and pressures were helping them to identify problems or issues earlier than in traditional testing processes followed by competitors. Specifically, the CFM56-2 variant was tested at 115 kN thrust, while it was certified at 107 kN and operated in the DC-8 aircraft type series at less than 98 kN of thrust [13]. In this way, CFMI was considering significant growth potential in their first family product, while also securing high engine reliability due to operation at much lower power settings than the designed ones. Furthermore, a 20% lower specific fuel consumption (SFC) was achieved compared to the established low by-pass ratio 10 ton turbofan engines of the period [12].

Next application for the CFM56 family was offered by Boeing with its new 737-300 aircraft fleet, leading to the development of the CFM56-3 variant. Less thrust was required for this aircraft type, but consideration of ground clearance effects was essential due to the low-wing arrangement. Consequently, side-positioning of the engine accessory gearbox and a smaller-diameter fan (60 in.) were deployed, resulting in lower propulsive efficiency compared to the CFM56-2 variant [12]. However, the lower operating range of fan corrected tip speeds offered significantly higher fan efficiency, yielding a 3% SFC improvement at cruise compared to the -2 variant [13]. A 90% commonality was achieved with the -2 variant, with the engine core being exactly the same as in the F101 engine. A considerable LPC redesign was necessary in order to cope with the new fan aerodynamics and ensure constant HPC flow capacity. Following the same strategy, the -3 variant was tested at 102 kN of thrust while certified and entered into service at 89 kN in 1984. Operated at 10–15% lower thrust levels than certified, led to the engine typically operating 150–200 °C below the certified exhaust gas temperature redline [13], with all the operational benefits that arose from that.

As can be seen from Fig. 2, F101, CFM56-2, and -3 variants lie on the same core technology levels, since the same core was utilized. It was not until the CFM56-5A engine variant that a “jump” in core technology was realized, targeting the 110 kN thrust class that would power the fly-by-wire A320 family. The main core technology advancements comprised: (i) HPC inlet guide vanes with smart stator schedules using the newly added full authority digital electronics control (FADEC), (ii) staged combustion due to lower fuel-to-air ratios, and (iii) improved HP/LP interturbine duct performance [12]. Aerodynamic improvements of all components using three-dimensional computational fluid dynamics were of major importance for this variant's core upgrade [15]. The upgraded core technology led to 11% better SFC than the -2 variant and the CFM56-5A1 entered into service at 110 kN of thrust in 1988 powering the fly-by-wire A320-211/311 aircraft.

The key feature of the CFM56-5A variant to provide substantial growth potential proved crucial in satisfying the upcoming market demands. The -5A core had sufficient margin to produce higher thrust either through simple throttle push, or by using an upgraded LP system. When the long-range Airbus A340 entered the market in 1991, the CFM56-5C solution was offered as a high-thrust derivative of the -5A1 for long-haul flights [16]. It was designed for 160 kN of thrust, incorporating a series of modifications on the LP system. A larger fan diameter of 72 in was required to provide sufficient air mass flow. Similarly, adding a fourth LPC (booster) stage and a fifth LPT stage provided enough power to the new high-flow fan [14]. Furthermore, a long mixed-flow nacelle was utilized [12]. Aerodynamic improvement of engine components, including core, was realized and a 5% SFC improvement was achieved. This introduced a new jump in core technology and established this core as the upgraded common one for the upcoming -5 variants. The -5C2 engine entered into service in 1993 at 140 kN thrust to power the A340-212/312 as depicted in Fig. 2. At that time, it was considered as the most powerful engine in the CFM56 family and the noise gold standard for an EIS 1995 engine [16]. Uprated -5C variants entered into service at 145 and 152 kN thrust as per market demands.

The upgraded -5 common core, the four-stage LPC and an updated 68-in fan design led to the development of the CFM56-5B variant, which was designed for small/medium range aircraft (180-190 PAX capacity). Each engine component downstream the LPC was consistent with the last -5A variant, including the thrust reverser [12]. The first application was the Airbus A321-112 aircraft, entering into service in 1995 and powered by the -5B2 engine model with 138 kN thrust. The CFM56-5B variant was the only engine that could power every model in the Airbus A320ceo family [16]. A performance improvement program had been fitted to the latest production configuration for the -5B variant, focusing on core and fan blades advancements [16].

The integrated upgrades of the last engine variant along with the already successful collaboration between CFMI and Boeing for the B737-300/400/500 aircraft, resulted in a new agreement; the CFM56-7B variant would now power the Next-Generation of B737 aircraft family. The same core technology as of -5B was utilized for the new variant, while fitting an advanced fan design of 61 in diameter [4]. The -7B variant would have a thrust range from 90 to 125 kN, meeting the requirements of the Next-Generation B737-600/700/800/900, which entered into service in 1998. Upgrade packages, including core and LPT advancements, had been fitted into the new CFM56-7BE model. They were fully interchangeable with other -7B engines and modules, and provided maximum flexibility to operators [16].

The development plan of the CFM56 engine program is depicted in Fig. 2. All engine variants are shown in Fig. 3 along with the aircraft type application. The latter figure is a proof of success for the CFM56 engine family. Common goals, common ambition, and the lack of competitive commercial engines between the partners were not the only reasons for this success. Providing high-performance and high-reliability products, as well as substantial growth potential of each common-core variant were of crucial importance to establish the CFM56 engine program as the one of the most successful engine families in the aviation history. This is proven by the fact that CFM56 engines power more than 30 different aircraft types for more than 600 operators across the globe [18], covering a thrust range from 85 to 152 kN thrust. By June 2019, the CFM56 fleet broke a new world record: it has become the first aircraft engine family in the aviation history to surpass one billion engine flight hours, carry more than 35 billion people, and fly more than 200 billion miles, which corresponds to 400 thousand round trips to the moon [23].

Over the years, the turboprop engine has been able to fill the aviation market gap between high cruise speed turbofan engines and low cruise speed piston engines. Conventional turboprop engines play a major role in the general, business, and regional aviation market due to high propulsive efficiency in the targeted speed range. Pratt & Whitney Canada's PT6 engine [24] is recognized as the most successful turboprop engine program worldwide, being the most widely sold of all turboprops [25]. The proven reliability and versatility of this engine family attracted the interest of the rotorcraft, and land-based power generation market, rendering it as the one of the most multipurpose engine families in world history.

The entire engine family power levels range from 360 to 1400 kW, keeping a common core throughout. The latter comprises a four- or five-stage axi-radial compressor, a reverse-flow annular combustor and a single-stage cooled or uncooled compressor turbine. A single- or two-stage free power turbine drives a propeller (for turboprop applications), or a rotor (for rotorcraft applications), or a generator (for land-based applications) through a single- or two-stage planetary epicyclic reduction gearbox [26]. The unconventional reverse flow arrangement chosen was the only solution enabling the integration of a free power turbine in such a small engine, since concentric shafts could not be implemented efficiently due to shaft diameter constraints [25]. Although doubtful in the beginning, this arrangement was proven highly successful. This was largely because velocities, and hence pressure losses, were kept low where considerable flow turning was required [26,27].

Each engine core component was carefully selected and designed specifically for this family, allowing power growth for future variants, if needed. The core cold section comprised three or four axial stages and a single centrifugal stage. This arrangement was preferred against an eight-stage axial compressor, since the latter would be prone to high tip clearance losses in the last stages, as well as vibration complexities [25]. As far as the hot section is concerned, a single-stage highly loaded compressor turbine was chosen. It was selected against a two-stage turbine, since the extra cost, weight, and general complexity outweighed the small performance gain associated with the lower stage-loading of the latter [25]. A single- or two-stage free power turbine was then decided subject to the desired power output levels. The free power turbine was coupled with a single-stage (gearing ratio of 5.33:1) or two-stage (gearing ratio of 5.33:1 and 2.82:1) reduction gearbox, depending on the turboprop or turboshaft application [27].

The outstanding effort of the design team in the 1950s to provide a simple, reliable, and flexible design with potential power growth for future applications was the main reason for the PT6 being established as the most successful turboprop engine in aviation history. As stated by Saravanamuttoo [25], the power growth of this engine family is primarily based on material upgrades and aerodynamic development of core components. That enabled higher compressor pressure ratios and mass flows, while only modest increases in turbine entry temperature took place. As highlighted by Badger et al. [28], a tripling of power output has been achieved through new technology infusion, while keeping the same engine diameter and adding only 25 cm to the original PT6-A6 engine length. Such growth enabled the potential of using this engine family for rotorcraft applications without fundamental conceptual changes [27]. It is also worth-noting that the PT6 layout offered the flexibility to install the engine in both types of nacelle configurations, puller (or tractor) and pusher, opening new markets in the aviation field. The wise choice to design for a rearward, reverse-flow inlet and forward facing turbine simplified the on-wing maintenance. Furthermore, it offered easy access to all external components, in stark contrast with the conventional competitor engines of the same class [24].

The PT6 engine family was introduced to the aviation world with the PT6-A small engine variant. It was a management decision that this engine family would focus on light engine applications. A market study was then carried out to explore the potential demand for aircraft engines at different power levels, and aircraft configurations over the following five-year term [29]. The development of the first model of this variant, the PT6-A6, commenced in 1958 with the testing of the engine core taking place in November 1959, almost 11 months after the initial design efforts [30]. A five-year development program led this engine model to enter the market in 1963, powering the Beechcraft King Air proof-of-concept model 87 at a power output level of 405 kW [24]. The specific feature of locating the two exhaust systems in the engine front (reverse flow arrangement) led to significantly shorter hot-section ducts within the engine cowling. As a result, the transition from piston engines to gas turbine engines was realized without requiring major modifications of the engine installation layout [30].

The initial market study demonstrated a need for gas turbine engines for both fixed-wing propeller and rotorcraft applications at a wide range of power levels [29]. The PT6-T “Twin-Pac” variant constituted one of the first PT6 family efforts toward medium-class rotorcraft propulsion in 1977, powering the rotorcraft fleet of the Italian Army, Navy, and Air Force [31]. The PT6-T6, being the first certified model of this variant, incorporated two PT6-A “small” turboshafts, coupled to a combining single-stage reduction gearbox with a novel clutch system. The latter permitted both singe and twin engine operation [24]. Medium-class rotorcraft (power levels from 1300 to 1500 kW) for oil exploration, emergency medical service, maritime patrol or utility operations constituted the major applications of this engine variant series. Electronic engine control had been added to the most recent models of this variant, while the turboshaft architectures were exactly the same as for the PT6-A model [32]. This engine variant established the company as one of the most successful engine manufacturers in the global rotorcraft market, with 11 developed engine models, 356 operators in 76 different countries, and counting more than 45 million flight hours [24].

Since its first model, the PT6 family opened a new market with increasing demand on a wide range of power levels for fixed-wing turboprop applications. The first core technology upgrade of the PT6 family had been recorded with the introduction of the PT6-A “medium” engine series, and specifically the PT6-A41 model with an EIS of 1977. A new compressor design with increased mass flow and pressure ratio had been incorporated, while retaining the same layout (three axial stages followed by a single centrifugal stage). Furthermore, substantial growth potential was provided by upgrading the core with vane cooling in the single-stage turbine, and by adding a second stage in the free power turbine [25]. A set of repair developments on the free power turbine and compressor turbine vane rings had been reported by Sourial [33,34] few years after the first PT6-A medium variant introduction.

The PT6-B variant was introduced in 1986 as a direct derivative engine of the well-performed PT6-T and the PT6-A medium engine series. It was a strategic movement to expand the company's reputation on providing powerful and versatile propulsion units for single- and twin-engine rotorcraft in the 700 kW power class [24]. As with each rotorcraft application, the two-stage epicyclic planetary reduction gearbox was replaced by a single-stage gearbox. As presented by Sweet [35], an engine acceleration schedule was developed for the PT6-B36B engine model, which was installed in the Sikorsky S-76B helicopter. The purpose was to provide improved engine responsiveness throughout the flight envelope while keeping adequate surge margin. An automatic fuel control and an electronic free power turbine governor were included in the most recent PT6-B37A model, powering the Leonardo AW119 helicopter [24]. A more optimistic application for the PT6-B35F model was the LearAvia Lear Fan 2100 turboprop business aircraft [36]. It had two engines in the back-end of the fuselage, coupled to a combining gearbox, in order to drive a tail-mounted constant-speed propeller. However, this application never entered production.

Demand expectations for a turboprop engine in the 900 kW class to power high-speed and high-altitude executive aircraft led to the development of the last PT6-A series upgrade in 1986—the PT6-A “large” variant. The option of a newly designed four-stage axial section coupled to the existing single centrifugal stage was chosen for this variant, to tackle the associated design risks and the desired up-rating potential [37]. As with former PT6 compressor components, the new design did not incorporate any variable guide/stator vanes, ensuring high reliability. However, an interstage bleed arrangement between the axial and the centrifugal compressor, as well as a jet flap system in the intake case were added. These design choices ensured surge-free operation during startup and idle conditions [37]. One of the assets added later on in this engine series was the power management system. It comprised: (i) the FADEC with advanced features to provide high levels of safety, (ii) the fuel metering unit, and (iii) the propeller interface unit [28]. This valuable addition offered pilots turbofan-like operation in a turboprop engine, by controlling the propeller blade angle automatically and scheduling the fuel flow based on chosen rating. Furthermore, a research program was carried out investigating the performance of the PT6-A65 combustion system using different fuels [38]. It showcased the potential impact on exhaust smoke and weak extinction characteristics on the associated combustors.

Continuing the legacy of the PT6-A large variant and the established success on the rotorcraft market, a new variant was introduced in 2003—the PT6-C series. It was mainly targeting the next-generation medium-class rotorcraft, including helicopter and tiltrotor aircraft. As highlighted by Smailys et al. [39], powering a tiltrotor aircraft, like the Agusta Westland AW609, entailed certain changes in the propulsion unit, compared to a helicopter- or turboprop-based propulsion system. The reduction gearbox was replaced by a high-speed direct-drive shaft, while a new single side-mounted exhaust system was incorporated. Running the engine core to higher operating temperatures and rotational speeds was essential to meet the increased power demand for these applications, while keeping the same core layout [40]. An improved combustor design with reduced CO and HC emissions featuring a new fuel nozzle system was integrated [39,41]. Electronic engine control with hydromechanical back-up and a dual channel FADEC were included in this engine variant to provide ease of maintenance and simpler pilot operations [24].

The latest core technology upgrade for the PT6 family at the time of writing can be found in the PT6-E variant. The (only) engine produced was the PT6-E67XP, powering the Pilatus PC-12NGX. It provides almost 900 kW of output power and gained certification in 2019. It features upgraded core technology with optimized turbine cooling and advanced single-crystal compressor turbine blades [24]. Optimized operation was achieved by integrated electronic propeller and engine control with digital propeller speed management. This enabled reduction in fuel consumption throughout the designated flight envelope.

Although the aforementioned applications of the PT6 engine core are aerospace-related, the program found similar success to other fields, further paying-off the initial investment costs for its development. In more detail, the PT6 engine core was the basis for the ST6 gas turbine, used in land-based and maritime applications. As reported by Closs and Robinson [42], the ST6 gas turbine was chosen to power a Norwegian-built high-speed boat with controllable-pitch propeller, and was also qualified for use by the U.S. Navy. However, several engine modifications were necessary to withstand the different operating conditions of maritime applications. Corrosion protection, compressor salt fouling, stalled power turbine capability, as well as the possibility to use a wide range of fuels were listed as the most important features [43,44]. The ST6-L813 engine, another aero-derivative version of the PT6 family, was used in the prototype MK 105 Airborne hydrofoil platform, providing electrical power for minesweeping operations [45].

The development plan of the PT6 engine program is illustrated in Fig. 4. All engine variants with the associated turboprop aircraft or rotorcraft applications are depicted in Fig. 5. The success of this engine family can be attributed to: (i) the sophisticated versatile engine design(s), (ii) the careful consideration of market gaps in the short- and long-term, (iii) the “embedded” power growth potential without need for core redesign, (iv) the continuous research and development efforts for product technology upgrade, and (v) the achievement of exceptionally high reliability in all engine variants.

As outlined by Wrong [27], the PT6 engine program is a great example of how, even for the most challenging engineering problems, the most elegant solution is perpetually the simplest one, and simplicity pays-off in reliability. The statistics of this engine family, as presented by Saabas in 2016 [55], reveal the great success of this program: 120 engine variants with 89 applications, flying for more than 400 million hours while experiencing a total in-flight shutdown rate of 3 per million flight hours.

The concept of growth engines originates from the generic engineering concept of product family. It was defined by Fellini et al. [56] as a set of products with similar architecture and common parts, while having different functional requirements. Product family was also defined by Meyer and Lehnerd [57] as a set of similar products that are derived from a common platform with specific features to meet different customer requirements. Several advantages of product family design have been reported in the literature [58]: reduced development risk and system complexity, improved ability to upgrade products at relatively short time, as well as enhanced flexibility of manufacturing processes. Each product variant or instance constitutes a member of the product family [59]. It should be highlighted that a product family aims at a predefined market segment itself, while each product variant targets subgroups of the market segment, meeting specific customer requirements [60].

The most dominant perspective for the definition of product platform has been reported by Meyer and Lehnerd [57], McGrath [61], and Robertson and Ulrich [62], who described it as a collection of assets—including knowledge, processes, and components—deployed in the development of a stream of derivative products. Simpson et al. [63] also defined product platform as a set of (common) parameters, features, and components that are fixed to a range of products within a given product family. The product platform was characterized by the following three features: modular architecture, interfaces, and rules to which all variants must adjust [64].

A product variety was defined by Ulrich [65] as the diversity of product variants provided to a predefined market segment over time. The question of “why to initiate and optimize a product variety” has been partially addressed by Ho and Tang [66]. They showed stimulation of sales and generation of additional revenue in case of product lines expansion and product differentiation. Further studies [67,68] have characterized this concept as a bridge to achieve the economy of scale, where cost of delivering variety is reduced by reusing proven knowledge on successful products [59]. It was claimed though [69] that the law of diminishing returns might prohibit further benefits from product variety. There is always a switch point between the customer satisfaction by the company's offerings in product variety and the level of complexity for the company that will keep the costs low. Thus, a variety management issue occurs.

The product family positioning problem should be also considered. It can be translated as “how to offer the right product variety to the right target market” [59]. Product variety by definition implies a wide range of products available for the customers, who are asked to choose. However, Huffman and Kahn [70] suggested that providing customers with a wide variety of products is wasteful and expensive, and might lead to mass confusion rather than customer satisfaction. This is why OEMs are forced to optimize product variety with respect to internal product complexity resulting from product differentiation [71].

But it is not only the optimization of product variety that an OEM should investigate. They have also to deal with the question of “when is the right time for each product variant to enter the market.” Several studies have revealed that demand for some product variants—although technically solid and reliable—is unexpectedly low. The reason lies on the diversity of customer preferences and time of product release. A mismatch of either reasons can cause a dramatic failure of product variant and, in some cases, of the entire product family. Studies on measuring customer preferences among many alternatives have been carried out by Green and Krieger [72]. They suggested that a conjoint analysis is the most suitable technique to initiate a new product family concept achieving customer satisfaction instead of confusion. Different weighting factors on each product variant released over time was proposed by Wassenaar and Chen [73]. In this way, the product family positioning problem in terms of time-frame entering into market can be tackled.

Strategywise, Simpson et al. [63] have reported two approaches on product family design: the top-down (a priori) and the bottom-up (a posteriori) approach. The top-down or a priori design approach is based on a company's strategic decision to initiate the development of a family of products based on a product platform and its derivatives. On the other hand, the bottom-up or a posteriori approach implies that a company redesigns or merges a set of already existing and separate products to standardize components and facilitate the economy of scale [63]. For the concept of growth engines discussed in this work, the a priori approach is dominant due to the company's initial strategic planning. However, part of the engine family design is reconsidered a posteriori based on product success and market evolution.

Another classification of product family design approach is the distinction between scalable and configurational. Du et al. [74] and Ulrich [65] explored the configurational or module-based design approach, where modules can be added, replaced, or removed to form new product variants across the family. On the other hand, the scalable product family design incorporates scaling factors or variables that “stretch” or “shrink” each product instance to meet different customer requirements [63]. Family of engines constitute one indicative application of the scalable approach on product family design.

A scalable product family design can be described as an optimization problem with common variables and scaling factors to be serving as design variables. As proposed by Nayak et al. [75], maximizing commonality and minimizing performance compromise of each product variant while satisfying different customer requirements are considered as the main objectives of the optimization problem. Du et al. [74] and Ulrich [65] suggested that maintaining high cost benefit across the product family should also serve as optimization objective, as a way to achieve the economy of scale. Similar studies on performance penalty analysis due to commonality have been carried out by Messac et al. [76] and Fellini et al. [56]. Both showed that optimization for the right combination of common and scaling variables is essential. Common variables can either be selected a priori [77] (can be named as parameters for this case) or be optimized along with the scaling factors [78,79].

A detailed literature survey on optimization approaches for the aforementioned engineering problem was performed by Simpson [80]. It was demonstrated that most optimization approaches follow the a priori method in order to limit the design space and reduce the computational burden. However, this comes at the expense of design freedom and optimal family performance, since designers must decide the common and unique variables within the product family before the optimization [81]. At that time though, there was limited knowledge on which variables affect the product performance the most. The literature survey carried out by Simpson [80] also shows that most of the approaches used multi-objective optimization considering three basic assumptions: (a) maximizing the performance of each product leads to maximum demand, (b) maximizing commonality among variants results in minimum production costs, and (c) the optimal product family is driven by modulating the tradeoff between the first two assumptions.

Treating a product family design as a multi-objective optimization problem requires the definition of evaluation metrics [82], which are based on three criteria: product commonality, product distinctiveness, and associated costs/revenues. Several metrics have been developed to measure and maximize commonality within a product family [83–85], while not surpassing a user-input performance penalty threshold [86]. Comparison of the various commonality indices to evaluate product families has been carried out by Thevenot and Simpson [87]. A metric to assess the individual performance of each variant is important to be defined as well. Within this context, Simpson et al. [81,88] proposed a product variety index which evaluates the tradeoff between commonality and individual performance compromises.

Furthermore, the associated costs and revenues should be estimated for each product family effort through an economic model. Relevant economic studies demonstrated that high commonality levels lead to reduced production costs but render product variants more indistinguishable from each other, affecting the product positioning in the market [89]. Several studies deployed economic models to capture the cost implications of a product family on top of the associated performance deviations [90,91]. The measure of profit or revenue, though, on a dollar-value base was considered unrealistic by Tarasewich and Nair [92], which led to the development of cost scale metrics [93–95].

Uncertainty consideration on single- and multi-objective optimization problems is common practice for challenging engineering tasks. The first effort toward a family of products with incorporated design uncertainty was recorded by Chang and Ward in 1995 [96]. Most of the listed optimization approaches by Simpson [80] had incorporated uncertainty in market demand and future product sales, as well as in customer requirements [97–101]. Technology shortcomings shall also be addressed as part of the uncertainty quantification, since the targeted component efficiency during the design phase might differ from the delivered efficiency at the end of the certification process. Considerating the aforementioned sources of uncertainty results in the development of a robust product family platform. Entering the first decade of 2000, many research efforts have been reported on robust design techniques for product family design in various applications—extending from aircraft design [102], to electric motors [63], and lift tables [103].

This section describes in detail a methodology framework for a product family design process which is tailored to gas turbine engines. The approach might vary from one engineering field to the other, but the framework skeleton remains the same and is based on the same principles. An illustration of the engine family design framework is depicted in Fig. 6.

Each engineering product, new concept idea or substantial market gap will initiate the discussion within a corporation's “Innovation and Strategy” department about potential product development or product family development. In all cases, market analysis constitutes an initial yet essential step for a successful product family development over the shorter and longer term. It forms Step 1 of the engine family design process, as depicted in Fig. 6.

As reported by Meyer [104], “a long-term success does not hinge on any single product.” It is the development of competitive and reliable series of products with growth potential targeting different market segments that enables the long-term success of corporations. As illustrated in Fig. 7 and presented by Meyer [104], there are four market segmentation strategies that a corporation can follow for a new product family development. The horizontal axis of each subfigure represents the different market segments that a corporation could potentially target for. Three qualitative segments have been selected for simplicity purposes. The vertical axis demonstrates the different levels of performance and price of the developed product variant and can be classified to low-, mid-, and high-level class. It is assumed herein that performance and price are proportional parameters.

Strategy I is called the niche-specific platform strategy, where each product is developed by its own platform, and focuses on a specific market segment. According to Meyer [104] and Simpson et al. [63], it is a common approach in industry, but it is not often observed in the gas turbine field. This strategy targets at individual product optimality at the expense of downstream costs [104]. Product commonality is not the focus and technology assets are not shared among the individual products.

Strategy II, called as horizontal leverage platform strategy, is characterized by “parallel scaling” for an established level of performance/price. This can take place either in the low-, mid-, or high-level class. A certain degree of shared technologies at component- and subsystem-level is realized across the targeted market segments, which accelerates product development and enables potential cost reduction. However, each product platform serves a wide variety of market segments and this can create risks in case of platform failure.

Strategy III features vertical leverage, where upward or downward scalability dominates the product development on the established market niches [104]. This strategy is based on the success of the initial platform, creating a customer desire for upgraded or downgraded common core products in terms of performance and price. Down-scaling a platform requires the removal of product design functionalities that will enable lower costs. On the other hand, up-scaling implies the addition of new technologies to meet higher functionality and performance targets. Shifting from high class to low class constitutes a weakness of this platform strategy, since downgrading the capabilities of an established platform is challenging. One major characteristic of this strategy is that it limits the scaled products to the same customer groups as for the initial platform, without expanding the targeted market segments.

Strategy IV, which is called as “the Beachhead platform strategy,” combines the horizontal leverage with up-scaling the derivative products [63,104]. A prerequisite for this method is an effective and successful initial product platform targeting at the lower class of performance/price. However, scaling-up is now targeting not only to the same market segment but also to new customer groups enabling an “upward diagonal move” in the market segmentation grid. This characteristic constitutes the critical difference with Strategy III, which targets only at the same market segment as the initial platform. A special feature of this method is the step-up scaling, where a jump from low class to high class product can be achieved through intermediate steps, depending on customer requirements and market demand. It is a fact, though, that a product which has been designed to be eligible for up-scaling with ease, will inevitably suffer a cost penalty in its original segment, when compared to the nonscalable products of the competition.

Strategy II represents the most common approach followed by engine manufacturers when initiating a new gas turbine engine family. A designated product platform is developed for the first engine variant design to enter the market. The same product platform will also be used by the following variant designs targeting different thrust classes or aircraft types. The latter enables the horizontal scaling. It is possible that there is no vertical axis in the gas turbine business. As per definition, vertical axis represents luxury but airlines purchase engines based on minimum direct operating costs.

During the development of the product platform, thrust ratings for the first engine variant but also for the entire engine family shall be foreseen. This is a very crucial feature, which will define the organization's niches in the market, along with its competitors. The lack of competitors on specified market niches can be described as market gaps, and it is the responsibility of the “Innovation and Strategy” department of each corporation to identify such opportunities. A successful strategy for initiating a common core engine program also includes the knowledge of which market niche each engine variant will target for. This phase is highly dependent on the current and projected customer demands, and requires the collaboration of different departments within an organization [4].

Several studies have reported that the success of an engine family with its derivatives is highly dependent on the foundation of common core technology and the available growth potential [57,61–63,104]. This statement introduces Step 2 of the engine family design process, as illustrated in Fig. 6.

It is important to decide if the initial engine architecture will remain fixed throughout the engine family or not. For propulsion system design, many literature studies have followed a fixed overall engine architecture throughout the engine family [4,5,9], but this is not always the case. The same engine core can be deployed for both military and civil purposes [3,56,105], serving as the core for turbojet, turbofan, and/or turboprop applications [8]. This can be proven by the utilization of the military F101 engine core in the development of the CFM56 civil engine family [11]. The number of applications increases as more design flexibility is offered to the design team to cope with operational and design limitations. But it is the potential balance between levels of design uncertainty and design flexibility that should be primarily taken into consideration.

Once the requirements for all engine variants have been defined, benchmark engine designs are next established for each application to serve as a comparison reference. Design freedom is allowed when establishing benchmark designs in order to find the individually optimized engine for each application. No (core) commonality rules are applied to these benchmark designs. Each common core variant will be compared with these individually optimized engine designs (that meet the same target requirements like thrust/power ratings), emphasizing on the resultant performance compromises, and cost benefits [4,105]. It should be mentioned that the comparison might take place against reference designs, but the decision on the engine family will be based against expected competition.

The step of conceptualization predefines the level of commonality for engine variants and it is primarily based on the design alternatives and restrictions that are imposed when formulating the engine family problem. Within the concept of growth engines, several design alternatives have been reported throughout the years. These can be summarized in Table 1 based on Sands [4]. But the list is expanded to include one more option, regarding the overall engine architecture. Specifically, an engine family program can be based on the following four foundations:

1. Core design. Either a geometrically fixed or a geometrically similar (or common) core can be employed. A geometrically fixed core is used when additional core flow is not necessary to meet the requirements or constraints for the variant applications and the core geometry is fixed. On the contrary, when more core power (through higher operating temperatures and/or core flow) is essential, a geometrically similar (or common) core can be used. Certain design rules must be followed to ensure core commonality, as presented in Sec. 1.2.

2. LP system design. Turbomachinery components connected to the LP spool are the focus for this design alternative. Their system design is always adjusted to the thrust or power requirements of each engine variant application. The design of these components can remain fixed or be fully redesigned without the need to follow commonality rules. In case of a turboprop application, the LPT is replaced by a free power turbine.

3. Technology levels. The engine core and LP system technology level can also vary among the engine variants. Component efficiencies and turbine metal temperatures constitute representative parameters to adjust technology levels, depending on the projected time-frame of EIS. This technology infusion might be applied to both the modified engine core (HPC, HPT, combustor) and the LP system (LPT, free power turbine, booster, fan, propeller) through performance improvement packages.

4. Overall engine architecture. Either a fixed or a modified layout can be implemented. Conservative engine family programs fix the overall engine architecture—to either turbofan only, turboprop only, or turbojet designs only—including the same number of shafts. There are studies, though, where a common core can be utilized across different engine architectures, as the case of the CFM56 engine family which was initiated by a military engine core.

The next step of the engine family selection process requires the formulation of the engine family design as a mathematical problem with clear assumptions, variable definitions, constraints, and targets. This is Step 3 of the engine family design process, as demonstrated in Fig. 6. A possible way to formulate an engine family design problem has been proposed by Hughes [106] in the form of a compromise decision support platform (DSP), but it shall not be taken as a definite direction. As per its original definition, DSP is a hybrid formulation of a multi-objective nonlinear optimization problem [107]. It follows a certain “logical” structure, which has been originally presented by Mistree et al. [108] and is illustrated in Table 2.

The scope is to find the values of system variables and deviation variables that minimize the deviation between actual system performance and system goals. The latter is an objective with the “right-hand side” being the target value or the aspiration level associated with the goal, and it is always expressed as an equality. It can also be considered as a minimum tolerable level of performance for the product variant. Examples of system goals for an engine family design can be levels of efficiency, fuel consumption, weight, or cost scales. Compressor exit corrected mass flow or compressor inlet mass flow corrected by pressure and temperature at exit conditions can serve as system variable for all engine variants. The feasible design solutions consider a number of system constraints (inequality in most cases) that should be satisfied; for example, a threshold in turbine metal temperatures, compressor last stage blade height, and compressor surge margin.

The compromise DSP was initially developed to provide support for human judgment in artifact design [108–110]. Hughes [106] reports that DSP cannot provide a real-world fully optimal solution, since it is not possible to consider all the necessary information for systems modeling correctly in the early stages of an engine family program. It does, however, provide much needed support to the designer for a good decision. Simulation using the compromise DSP have been carried out for ship family design [111], aircraft family design [102,112], electric motor family design [63], as well as thermal energy systems [113].

The fact that the suggested DSP cannot take into account all the information early in the engine family design, sets the foundation for uncertainty consideration. Thus, the above approach can be formulated as a robust design optimization problem.

The next step of the engine family design process, Step 4 in Fig. 6, includes the selection of appropriate simulation tools. For the growth engine concept, engine cycle performance, and conceptual design tools should be deployed. The numerical propulsion system simulation developed by NASA [114], GasTurb initially developed by Kurzke [115] and now owned by GasTurb GmbH (116), the environmental assessment framework developed by Kyprianidis [117] and the propulsion object oriented simulation software (proosis) initially developed by Alexiou et al. (118) and now integrated within EcosimPro (119) constitute established and validated tools to carry out engine performance assessments and conceptual design studies for any conventional and novel propulsion system architecture.

These tools should be integrated in a greater platform (be it compromise DSP or similar) to realize the design of benchmark engines and engine variants for the designated family. Benchmark engines (or in other words, individually optimized engines without the need for core commonality) should be designed for each targeted market niche and will serve as comparison reference against each engine variant (or in other words, engine designs with a certain level of core commonality). In case of computationally expensive simulation, building and validating metamodels [63,120], using response surface and Kriging [121], is highly recommended in order to reduce the computational burden. At the end of this step, a feasible design space of potential engine variants is created for each market niche, and potential tradeoffs are identified when compared against the benchmark references.

The evaluation of engine variants against benchmark designs forms Step 5 in the engine family design process. The comparison should be consistent, and the design decisions should obey to the same set of assumptions for both engine design philosophies. The philosophy of benchmark engines focuses on finding the individually optimal engine without considering any form of commonality. On the other hand, the design philosophy of engine variants is primarily based on core commonality, that will compromise individual performance but at the same time will guarantee rapid and competitive engine development for different market niches over time, distributing the massive investment cost across several applications. Metrics, or figures of merit, have to be selected in a careful manner, in order to highlight any performance compromise between these design philosophies.

Quantifying the performance compromise and potential tradeoffs between benchmark designs and engine variants require the introduction of two indices, as suggested by Simpson et al. [88]: non-commonality index (NCI) and performance deviation index (PDI). They are considered as the most dominant metrics to evaluate the feasibility of an engine family design.

where i is the current application for common core engine designs, N_apps is the total number of applications for different market niches that the engine family targets for, w_i is the weighting factor used for each application, x_cm is the common value of the system variable for all applications, and $xi*$ is the value of the same system variable for the individually optimal engine of each application.

where y_i is the value of the system performance parameter for each application and $yi*$ is the value of the system performance parameter for the individually optimal engine of each application, while i, N_apps and w_i have the same definition as above.

The system performance deviation and commonality for two applications are depicted in Fig. 8, as originally suggested by Fellini et al. [122] and further demonstrated by Sands [4]. Only two indicative applications, A and B, are represented in the figure for the sake of simplicity. HPC exit corrected flow or HPC inlet mass flow corrected by pressure and temperature at exit conditions constitute the dominant system variables for common-core variant studies [3], used in the abscissa of Fig. 8, while SFC, specific core power, overall efficiency, or total weight [4] can represent the ordinate. The chosen system variable has a certain effect on the selected performance metric (SFC, specific core power, overall efficiency, or total weight) of each application, depicted here with arbitrary concave curves of different size. The individually optimal designs for each application are illustrated by hollow dot points with coordinates $(xA*,yA*)$ and $(xB*,yB*)$⁠. These two designs can also be considered as the benchmark designs.

The gray-shaded area represents the feasible common design space. The chosen system variable, x_cm, can take any value on the x-axis in the range $[xcm1,xcm2]$⁠. This selection will yield certain engine performance levels for application A and B, described by y_A and y_B, respectively, and defined by the gray vertical solid line. The value of x_cm is selected by the design team to be used for both applications. This means that the selected compressor exit corrected mass flow or compressor inlet mass flow corrected by pressure and temperature at exit conditions is chosen to serve both applications A and B. The associated metrics of PDI and NCI for the selection of this x_cm are also depicted in the figure. Specifically, performance compromise or penalty for the common core design selection is demonstrated by the vertical difference in performance targets $(yA*−yA/yA*,yB*−yB/yB*)$⁠, marked as PDI_A and PDI_B. The commonality deviation can be shown by the normalized horizontal difference on system variables $(|xA*−xcm|/xA*,|xB*−xcm|/xB*)$⁠, marked as NCI_A and NCI_B.

The behavior of both PDI and NCI metrics within the feasible common design space (gray-shaded area) for different values of $xcm∈[xcm1,xcm2]$ form in essence typical Pareto fronts. Their shapes are depicted in top right part of Fig. 8. Moving the value of x_cm from $xcm1$ toward $xcm2$⁠, the performance penalty for application A (PDI_A) increases, while the performance penalty for application B (PDI_B) decreases. The same trend is observed for the non-commonality index levels, NCI_A and NCI_B.

As noted before, w_i represents the weighting factors for each application. It is usually common to apply increased weighting factors to applications where less compromised performance is desired. This can be either the first variant application of the engine family to avoid a premature engine program failure, or an engine variant application with high confidence of popularity [4]. The evaluation process has to ensure that common core designs will be robust to future customer requirements and constraints, as well as to projected technology fusions. In this way, the initial design will provide competitive solutions for their customers both now and in the distant future.

Envisioning the future of aviation, it is widely believed that gas turbine engines will still play a major role in aircraft propulsion for the next decades, especially for medium- and long-haul aircraft classes [123,124]. Within this context, electrification and novel (hybrid) propulsion system architectures have emerged as promising concepts to tackle the tight environmental targets set by advisory and regulatory bodies [125–127]. Significant battery energy density improvement is required, though, to enable propulsion systems with reduced SFC and higher efficiency levels.

As with conventional configurations, hybrid-electric engine architectures can be designed considering the concept of family of engines (or growth engine concept or core commonality) early in the design process. However, many more components will be involved in this principle of propulsion system design when compared to the conventional growth engine concept. The reason lies on the fact that electrification entails not only the use of gas turbine engines but also electrically driven propulsors, electrical power systems, energy storage, and thermal management systems.

Applying the principles of the growth engine concept to the design of hybrid-electric aircraft propulsion systems is expected to attract the attention of the aviation industry in the upcoming years. Transitioning, though, toward the concept of growth electrified propulsion systems requires the deep understanding of:

1. fundamental engineering theory of product families and its application to different fields, as described in Sec. 3,

2. the growth engine concept as developed for conventional engine configurations, as presented in Sec. 4, and

3. the principles and challenges of electrified propulsion systems.

The engine family design process, which is presented in Fig. 6 for the conventional growth engine concept, requires certain modifications to allow the transition toward electrified architectures:

1. Electrification provides an additional degree-of-freedom in the design process when compared to conventional engines, and should be considered during the engine family conceptualization (Step 2 in Fig. 6). This gives one more design alternative, namely, “Electrification Scenario,” that should be added as a fifth row in Table 1. The choices for this alternative can be either conventional (pointing to configurations where fuel is the only energy source) or hybrid-electric (pointing to configurations where both fuel and battery are used as energy sources). The hybrid-electric choice can be further split into parallel hybrid, series hybrid, or series/parallel partial hybrid.

2. Design assumptions, choices, and constraints for each subsystem in a hybrid-electric architecture (gas turbine engine, additional propulsors, thermal management system, and electrical power system) shall be taken into consideration during the problem formulation (Step 3 in Fig. 6). Assumptions for battery energy/power density levels, the design choice for the degree of hybridization, the selection of electric machine topology, as well as electrical power system constraints for battery cell and motor peak temperature levels constitute basic modifications of Step 3 during the electrified family design process.

3. Electrified benchmark engines (in other words, the individually optimized electrified propulsion systems without the need of core commonality) should be also designed during Step 4 of the process, along with the conventional benchmark engines (individually optimized non-electrified engines without the need of core commonality). The evaluation that will follow in Step 5 using the metrics of NCI and PDI should compare both types of benchmark engines with the common core variants of the engine family.

A representative paradigm for the concept of growth electrified propulsion systems is depicted in Fig. 9. It forms a scenario that could simply explain the rationale behind the proposed concept. The ordinate of this figure represents the “Engine Architecture,” as presented in Table 1. The abscissa illustrates some choices of the additional variable of Electrification Scenario, namely, conventional, parallel hybrid-electric, and series/parallel partial hybrid-electric. The remaining design alternative variables presented in Table 1 are “hidden” behind each propulsion system variant.

A family of electrified propulsion systems with common core design can actually start from a conventional gas turbine engine that will define the core design. During the core defining engine design, all Steps 1–6 presented in Fig. 6 should be followed. Market analysis, potential market niches, and initial engine design requirements should be taken into careful consideration when defining the engine core. Once the initial propulsion system design is realized, conceptualization of upcoming variant scenarios can follow, just like in Step 2 of Fig. 6. Keeping a consistent architecture for the entire family (be it turbofan, turboprop, or turbojet) but with a different electrification scenario (conventional, parallel, series, or series/parallel partial hybrid) for each engine variant is possible.

Moving from a conventional turbofan engine to a parallel hybrid turbofan engine constitutes a low-risk core commonality strategy. It is reported that engine operability and SFC are benefited from electrical power input at design point through the LP spool (booster or IPC) [128], while keeping the engine core unchanged. In such parallel hybrid systems, there can be bidirectional power flow (electrical power in-take from the battery to the LP spool and electrical power off-take from the LP spool to the battery) for each flight phase to improve engine operation accordingly. Engine core can either remain geometrically fixed but operating at different ratings (temperature and pressure levels), or be modified according to the engine variant design rules (as presented in Sec. 1.2). In both cases, the LP system can be modified or be completely redesigned to meet the thrust requirements depending on the electrical power balance from the battery to the LP spool (battery charging or discharging). The electrical power system, the energy storage, and the thermal management system are also part of the growth electrified propulsion system. However, this work focuses only on engine design and hence, the application of product family principles on the rest subsystems will not be explored.

The series/parallel partial hybrid propulsion system forms an alternative to apply the common core concept on electrified architectures. This electrification scenario requires the presence of additional propulsor(s), which enables thrust split between the electrified common core engine and the additional propulsors, leading to substantial common core engine down-sizing. For example, the core of the core-defining engine (let's assume it is a conventional engine without electrification) should be designed having in mind that it will be used in parallel hybrid and series/parallel partial hybrid propulsion systems, as illustrated in Fig. 9. That would imply that the electrified common core engine is running at different thrust levels, since the electric power input from the battery and the presence of additional propulsors enable engine derating (thrust split). The LP system is more likely to be completely redesigned for the new thrust levels, but the core can be designed at first hand in a way that it remains geometrically fixed or geometrically similar for all applications. Core upgrade while retaining core similarity can be achieved by compressor flaring or zero-staging, resulting in core mass flow up-scaling up to 20% [4]. On the other hand, it is also possible that the core compressor of one engine variant forms part or a subset of the original core compressor. A representative example constitutes the HPC design strategy for the E3E turbofan engine and the TP400 turboprop engine, as described by Gümmer et al. [129]. The E3E HPC was designed with nine stages in total, having a corrected inlet mass flow requirement of 30 kg/s. The TP400 HPC had a requirement of 12 kg/s of corrected inlet mass flow and thus, was formed by the inner block of stages 3–8 of the E3E compressor.

The family of electrified propulsion systems can be extended to other alternative design directions as well. For example, the common core design for a turbofan application can be applied to different engine layouts (turboprop or turbojet). Figure 9 demonstrates a case of common core re-usability between turbofan and turboprop architectures. The electrification scenario can also vary for the turboprop-driven propulsion system family in the same manner as discussed for the turbofan (parallel hybrid and series/parallel partial hybrid).

It is important to keep in mind that a concept of growth (electrified) propulsion systems is always associated with performance penalties for variant designs, when compared to individually optimized (benchmark) solutions. However, there is a high potential for better economics of the entire family in light of a systematic design process and planning.

Designing and developing a clean-sheet aero-engine can cost several billions of dollars, and is typically initiated once in ten years. The electrification potential and strict regulations for more energy-efficient air transportation have reinforced the need for innovative yet economically viable propulsion systems to enter into service. Within this context, the concept of using a geometrically fixed or geometrically similar engine core across a family of engines can be considered as a promising solution. The aforementioned concept of growth engines is based on the generic engineering paradigm of product families. The main benefit is the rapid development of engine variants and the distribution of the massive initial investment of an engine program across a number of aircraft applications. However, the major challenge of this concept lies on the risk of having a non-optimal core for each engine variant and its potential competition against individually optimized engines in the targeted market niches.

Bearing this concept idea in mind, a comprehensive literature review has been presented, focusing on the motivation, the fundamental theoretical background and the implementation methodology of product families. Gas turbine engine (conventional or electrified) has been the product under discussion. According to literature, the growth engine concept originates from the generic engineering principle of product family, which has been initiated and applied many years ago to several engineering fields. Engine family positioning and the timeframe that each variant will enter the targeted market constitute major challenges of this concept. However, the design and development of engine variants with substantial growth potential can serve as a bridge to achieve the economy of scale, enabling the long-term success of corporations. It has been highlighted that the concept of growth engine is always associated with performance compromise of individual variants but potentially better economics for the entire engine program.

This review work starts with the analysis of two realistic case studies in order to showcase the benefits and implementation of the growth engine concept over the years. Two of the most successful engine programs in aviation history for turbofan and turboprop/turboshaft layouts have been presented, namely, the CFM56 and the PT6 engine families. The lessons learnt from both programs can be summarized in two points: (i) thorough market analysis and identification of potential market gaps in the short- and long-term timeframes and (ii) sophisticated versatile engine designs considering the fact that simplicity pays-off in reliability.

The current work has also presented an implementation methodology to design and evaluate a gas turbine engine family based on the open literature. Five steps are included in the engine family selection process. Establishing a market segmentation strategy and identifying market gaps are crucial for the initial engine family positioning and the potential success of the program. Engine family conceptualization is carried out afterward, where design variables, constraints, and goals are defined along with the establishment of benchmark engines (in other words, individually optimized engines without the need for core commonality). Furthermore, a compromise Decision Support Platform (or any similar integration platform) can be used to formulate the product family design problem accounting for possible uncertainties on design requirements, objectives, and market demand. Sophisticated, robust and preferably rapid engine design simulation tools have to be integrated within the platform to realize computationally-efficient designs of engine variants. Two stages of evaluation are presented for the final step. The first stage targets the assessment of each engine variant against the individually optimized benchmark engines. The second stage focuses on a holistic engine family assessment, identifying tradeoffs between commonality and performance compromise. The selection of the most promising engine family is supported by a human-machine interface that helps engineers to take the final decision.

This review work ends with an effort to shed light on the implementation of the growth engine concept to the design of hybrid electric propulsion systems. Certain modifications are required in the methodology, the design principles and design alternatives to account for the additional subsystems that are present in hybrid-electric architectures (e.g., electrically driven propulsors, electrical power system, energy storage, and thermal management system). Although these subsystems can provide an extra degree-of-freedom in the design of the gas turbine engine, significant design constraints can still limit the design space of potential families. However, similar principle-of-practice can be applied as for the conventional growth engine concept. The major criterion would still be the distribution of the massive initial investment.

Literature has shown that the concept of growth engines in conventional applications still has room for improvement in terms of optimization techniques and robust design methods. This offers many opportunities to systematically investigate and further extend this concept to novel technologies in aviation. A roadmap of transitioning the concept of growth engines from conventional applications to electrified ones over time is illustrated in Fig. 10, revealing the research questions associated with each crucial step along the way.

Moving toward the electrified analogy, the inclusion of additional subsystems implies the deep understanding of each component's principle of operation and associated design constraints. A major benefit stemming from electrification is the flexibility of powertrain design. The level of flexibility depends on the selected electrification scenario. Parallel hybrid and series/parallel partial hybrid architectures provide substantial capacity to implement the concept of growth propulsion systems. The presence of primary and secondary energy sources (chemical and electrical energy) in these architectures enables a high number of power management strategy combinations. The first research question (RQ1) therefore rises:

Technological advancements in the associated subsystems play a crucial role on the final design of the electrified engine family, allowing for substantial core growth potential. Given that the electrified aircraft propulsion system designs are largely conceptual at this time, many components lack of technology maturity and limit the design space of feasible solutions. However, this should not be a barrier for the computational implementation of this concept. On the contrary, it might enable new directions of technology improvement packages on “bottleneck” components. Development of high specific energy battery and high specific power electrical machines are necessary. Compliance with airworthiness and certification requirements as well as top level aircraft requirements are necessary. Technology advancements on materials for electrical machines, power electronics, and turbomachinery components can shorten the way toward electrification. Waste-heat exploitation for aircraft-related purposes and use of new cooling mediums present opportunities to improve the technology of thermal management systems. The second research question (RQ2) therefore emerges:

Interdisciplinary computational frameworks are necessary to provide feasible solutions for such complex engineering problems. Physics-based models, optimization algorithms, probabilistic methods, and objective evaluation processes shall be combined for a successful and meaningful outcome. A human-machine interface is required to further interpret and assess the results. Versatile engine conceptual design tools are needed, which are validated against experimental data and high-fidelity computations. Uncertainty at different stages along the process (market demands, customer requirements, technology shortcomings, and design assumptions) should be quantified in order to provide robust solutions with certain confidence interval. Otherwise, deterministic solutions might have direct cost implications on the interested corporation. Robust design optimization algorithms provide reliable solutions on complex engineering problems. The evaluation of an electrified family of engines, though, requires metrics for fair comparison against conventional counterparts during the final selection. Design implications like performance compromise, cost benefit and design commonality should be accounted together with the implications brought up by electrification itself. The third research question (RQ3) therefore rises:

The authors are grateful to Professor Anestis Kalfas from the Aristotle University of Thessaloniki in Greece and Dr. Ioanna Aslanidou from Mälardalen University in Sweden, for their insightful advice and continuing support. The authors would also like to acknowledge Mr Dimitrios Bermperis for the stimulating discussions on advances in electrical power systems for future propulsion concepts. Many thanks go to the reviewers of this work for their constructive feedback to improve the overall quality and clarity of the article.

• This work was performed within the project HECARRUS, which has received funding from Clean Sky 2 Joint Undertaking (JU) under the European Union's Horizon 2020 Research and Innovation Programme with Grant Agreement No. 865089 (Funder ID: 10.13039/100010661).

The authors attest that all data for this study are included in the paper.

CFMI =

CFM international

DSP =

decision support platform

EIS =

entry into service

FADEC =

full authority digital electronics control

GE =

General Electric

HP =

high pressure

HPC =

high pressure compressor

HPT =

high pressure turbine

IP =

intermediate pressure

IPC =

intermediate pressure compressor

LP =

low pressure

LPC =

low pressure compressor

LPT =

low pressure turbine

NCI =

noncommonality index

OEM =

original equipment manufacturer

PDI =

performance deviation index

RQ =

research question

SFC =

specific fuel consumption

T3 =

compressor exit temperature

T4 =

combustor outlet temperature