6. Maintenance Cost Estimation
6.5.1 Maintenance Cost Model Selection
A detailed workscope for engine maintenance will naturally be the most accurate predictor of engine maintenance cost. However, this form of analysis is more applicable to in-service engines with a known operating procedures, known part cost, and requisite maintenance activ-ities. Maintenance cost per year or flight cycle is then the result of a detailed schedule of work for engine maintenance. This detailed analysis is not useful for a preliminary cost estimate, and parametric cost estimating relationships are once again the most feasible option. In addition, detailed component design is not available for a preliminary design configuration. This rules out component-based maintenance cost estimates. Although a detailed maintenance cost estimate is not feasible, it is still important for the model to be able to represent maintenance cost as a function of engine operation. As identified in the previous section, maintenance requirements are dependent on factors such as local environment (e.g. hot airports) and flight cycle length.
Flight cycle length is a particularly important aspect that must be represented if the cost of missions other than the baseline design mission is to be simulated. For a novel aircraft, main-tenance cost cannot be estimated by scaling a known baseline cost, as there is no data from which a baseline value can be obtained. Applicable methods must therefore rely on regression analyses to identify how maintenance cost is related to known engine design variables such as thrust and temperature.
A selection of maintenance cost estimating relationships were chosen for the propulsion system and airframe. The combination of models provides a wider range of values for com-parative purposes. Of the selected models, four covered the maintenance cost of engines and two covered the maintenance cost of the airframe. The following models were used to provide estimates from a range of time periods:
• ATA, 1967 (Airframe and engine models) [188]
• Liebeck et al., 1995 (Airframe and engine models) [54]
• Kang et al., 2008 (Engine model) [186]
Table 6.4: Inputs for the selected cost estimating relationships ATA Liebeck Kang Seemann
Airframe
Airframe Weight X X -
-Block Speed X -
-Aircraft Cost X -
-Flight Hours (FH) X X -
-Engine
SLS Thrust X X X
Engine Weight X
Engine Cost X
Component Performance X
Component Stages X
Component Diameters X
Engine Flight Hours (EFH) X X X X
• Seemann et al., 2011 (Engine model) [191]
The cost estimating relationships presented by ATA, and Liebeck et al. break costs down into labour, materials, and maintenance burden to create an overall cost per trip or cost per flight hour. Separate relationships are provided for the airframe and engine. The relationships pre-sented by Kang et al. and Seemann et al. estimate maintenance cost by predicting the interval between shop visits, the shop visit cost, and the cost of life-limited parts. Subsequently, these values can be used to estimate the average cost per flying hour or the cost per trip. The rela-tionships provided by Kang et al. use the performance and dimensions of the fan, compressors and turbines during SLS operation (temperatures, pressures, tip speeds, number of stages and diameter) to create a cost estimate. An estimate of the engine dimensions is therefore required.
This was obtained using in-house software for gas turbine sizing (Section 3.4.1). The model by Seemann et al. relies on a simpler relationship linking maintenance cost to the engine’s thrust and weight. Table 6.4 breaks down the inputs required for each of the selected cost estimating relationships. None of the models provide standard errors of estimate. Therefore, confidence intervals cannot be produced for the maintenance cost estimates. The influence of flight time on the engine maintenance cost per flight hour may be presented as on a maintenance cost severity curve [182]. As each model includes the number of flight hours as an input, they may be used to produce severity curves similar to the example shown in Figure 6.17 by varying the input flight length.
It was assumed that a novel airframe configuration would not have significantly higher main-tenance costs than a conventional planform, and hence that current models would be able to create a reasonable estimate. However, the selected models do not account for the poten-tially higher maintenance cost attributable to the use of a composite structure. The airframe maintenance cost models are predominantly a function of airframe weight. Therefore, a lighter airframe would lead to a lower cost estimate, even where a lighter airframe is the result of us-ing advanced materials. Section 6.4 identifies that the RAND acquisition cost model includes materials weighting factors to scale the estimate based on the use of advanced materials.
These materials weighting factors account for the fact that low weight can be the result of more expensive or novel materials. It is useful to incorporate a similar functionality in the airframe maintenance cost models. Both airframe maintenance cost models break the cost estimate into comparable groupings to the acquisition cost estimate: labour and materials. It was therefore be assumed that similar materials weighting factors may be applicable for the maintenance cost
6. Maintenance Cost Estimation
models. As an initial estimate, the material weighting factors from the acquisition cost modelling section were used. The two relevant factors are the materials and labour scale factors for air-frame manufacture from Resetar et al. [173] and Younossi et al. [174]. The baseline aircraft is manufactured primarily from aluminium, with a small percentage of composite materials. The resultant material cost factors for the baseline aircraft therefore lead to a reduction in mainte-nance labour cost (WMCF = 0.932) and an increase in material cost (WMCF = 1.039). These materials weighting factors were used to provide an initial estimate of the influence of advanced materials on the cost estimate.
Maintenance Cost for the N3-X
Predicting maintenance cost for a future novel aircraft is one of the main challenges in the techno-economic analysis. The selected engine maintenance cost models are derived from data for conventional turbojet and turbofan configurations. It can therefore be assumed that a reasonable estimate would be produced for the maintenance cost of the baseline aircraft’s engines. However, these models will becomes less applicable as propulsion system configura-tions diverge further from current engines and as new components are added. This is especially true for the turbo-electric propulsion system of the N3-X, which incorporates a superconducting electrical system and multiple propulsors, rather than two to four individual propulsion units.
As a result, the engine maintenance cost models are less applicable to the novel propulsion system configuration of the N3-X.
However, it can be assumed that a cost estimate could be produced for the turbomachinery, as these would likely be a similar configuration to more conventional propulsion system. As the main engines are closer to turbogenerators than turbojets, they do not produce a thrust term suitable for the cost estimating relationships by Liebeck et al., Seemann et al., and Kang et al.. As a preliminary estimate, the equivalent thrust-producing engine could therefore be used instead to represent the turbogenerators for a maintenance cost estimate. Whilst industrial gas turbine models may also be applicable for a power-producing engine, aero gas turbine models were used to ensure a consistent estimate between thrust-producing and non thrust-producing variants of the N3-X main engines.
The remaining two sub-systems are more challenging when creating a cost estimate. Su-perconducting electrical machinery is not currently used in commercial aircraft applications. At the time of the study, there is therefore no information available in literature on predicting its maintenance cost. In addition, industrial electric machines are at a difficult scale to electrical machines for an aircraft. Industrial maintenance cost models were therefore concluded to be unsuitable for the purposes of this research. Predicting the maintenance cost of an array of dis-tributed fans is equally difficult, as such a configuration is not currently used on aircraft. Of the selected models, the maintenance cost for a fan is generally encompassed within the overall maintenance cost estimate for a turbofan engine.
These electrical system and propulsor array are key subsystems in the propulsion system.
However, no information is available at this point in time to create a maintenance cost estimate for these two subsystems. Creating a maintenance cost estimate for the turbomachinery alone would therefore lead to an estimate that is likely to be unreasonably low, as it would neglect a large portion of the propulsion system. In order to obtain a preliminary cost estimate accounting for the whole system, the thrust and weight of the propulsion system as a whole was used (i.e.
including the electrical system and propulsor array), as opposed to the thrust and weight of the equivalent engine. An alternative assumption may be to assume that the propulsor array needs minimal maintenance, as it simply consists of a number of fans. However, an additional cost model would still be required to predict the maintenance cost of the electrical system.
Therefore, given the scope of the study and the modelling fidelity requirement, this assumption is believed to be suitable. However, further work may wish to develop cost estimates for a turbo-electric propulsion system.