Cost estimation in an aeronautical Supply Chain T. Murino1, G. Naviglio1, and E. Romano1 1
Department of Materials Engineering and Operations Management University of Naples “Federico II” P.le Tecchio 80 – 80125 Napoli ITALY The Customer Support in an aeronautical supply chain is the key element, present throughout the cycle life of the aircraft for customer loyalty and ensure a business long term. Every order is different since each customer requires that their aircrafts respect particular specifications, then it is necessary to carry out an estimation cost of spare parts Service each time there is a certain request. At present the calculation process used in order to define The Spare Parts Service is an “Exact Estimation” method, through the OPUS10 Tool, very accurate, but extremely expensive in terms of time and processing time. Since, unfortunately, not all orders come to a successful conclusion, in the first negotiation phase (RFQ) it is possible to use an approximated calculation method allowing a great time and thus costs reduction. This paper goal is to search a mathematical algorithm, in a necessary approximation circle, that can substitute an exact method with a parametric estimation methodology. Index Terms— CRM, non-repairable and repairable systems, aeronautical supply chain, spare parts estimation, customer support costs.
I. INTRODUCTION In an aeronautical Supply Chain, the aircraft purchasing is only the first stage in the customer relationship process. Operations and logistics support is provided by a dedicated business unit as a “Key in Hand Service” in order to rationalize needs in this sector and to optimize the balance between costs and benefits. An effective logistics support must be an independent product. The customer support controls any phase of supply chain, it provides necessary tools and it assures the aircraft’s maintenance. It includes also the components removal and repair with client firms until the product is restored.. Therefore, the Logistics Support is the key element presents during the entire Aircraft Life-Cycle in order to make a client a faithful customer and to assure a business for a long term. The “Customer Support”, or commonly “Post Sale Service”, represents the main tool able to establish the link between finite product and customer’s needs. An efficient organization dedicated to “Customer Support”, integrated in the company system, provides customized solutions aimed to maximize the aircraft’s availability and to optimize the “Through Life Support” costs. An high engineering level, with a promptness of answer and an efficient Supply Chain control and an effective customer relationship management, are the main elements in order to assure customer satisfaction and loyalty. Supporting the product and the client means to answer customer’ s need, defining the support factor, giving a support and handling the product maintenance. In this paper we have analyzed the cost estimation of the spare parts in the aeronautic field and in particular the aim of this work is the definition of an algorithm that allows a parametric estimation of spare parts costs with less resources and time. II. THE CUSTOMER SUPPORT IN AN AERONAUTICAL SUPPLY Digital Object Identifier inserted by IEEE
978-1-4673-0248-7/11/$26.00 ©2011 IEEE
CHAIN
The customer support in the aeronautical field is a range of services providing assistance in order to maximize the availability of the aircrafts and to optimize the costs, as showed in the following figure 1.
FIG. 1 The customer support services The customer support has, also, an important weight on the total cost of a order. In particular the spare parts service is the most important service of the customer service because it affects the 60 % of the customer supports costs (fig. 2).
FIG. 2 The customer support costs So the estimation of the spare parts costs means to calculate the customer support costs. The Customer Support is the key element, present
throughout the cycle life of the aircraft for customer loyalty and ensure a business long term. The current commercial process with the client always passes through an early stage, called RFQ (Request for Quotation), from which is sufficient an “estimate of costs”, and a more advanced stage, called RFP (request for Proposal) for which is necessary a “detailed assessment”. Every order is different since each customer requires that their aircrafts respect particular specifications, then it is necessary to carry out an estimation cost of spare parts Service each time there is a certain request both RFQ or RFP. In both cases the results coming from an accurate estimation are very expensive. III. THE SPARE PARTS COST ESTIMATION Currently, the methodology used to calculate cost of the customer support is very accurate but requires a lot of workload and performance time (Accurate Estimate). As a matter of fact, several data such as MTBF, maintenance policies, lead time, number of operational bases, number of aircrafts, an so forth are used as input in a software called OPUS10, which returns a quote after some simulations. It ends only after ten days, with considerable use of resources and thus costs (fig. 3).
FIG. 3 The accurate estimation of the customer support costs Since, not all business negotiations are successful so early in the RFQ stage (Request for Proposal) would be useful an algorithm that allows us to reduce response times and costs (fig. 4). This paper goal is to search a mathematical algorithm, in a necessary approximation circle, that can substitute an exact method with a “Parametric Estimation” methodology.
FIG. 4 The accurate estimation vs parametric estimation
The mathematical algorithm that will replace accurate estimation method will be implemented through a easy tool such as Excel. This experimental methodology used in order to define the algorithm provide different steps. The first step is to reduce the variables number, "freezing" a logistical scenario. Secondly, some simulations run with the accurate estimation software (OPUS 10). The last step is to find the envelope of all simulations in order to obtain an algorithm allowing to get a parametric estimation. A. Logistic scenario definition The first step is to define the logistic scenario, so it is possible divide the parameters in two categories: constant parameters such as stations, maintenance policy, MTTR, lead times, scheduled maintenance and overhauls, transport times, storage policy, life support, reliability data, cost of parts; and variable parameters such as fleet size, fleet usage, fleet availability. B. Simulations with the accurate estimation software The simulation software requires the division in repairable, non repairable and consumable components. The non-repairable systems are the ones that are not repaired when they fail. It doesn’t necessarily mean that they cannot be repaired, rather that it isn’t economically convenient since the repair will cost as much as the purchase of the new unit. In their turn the non-repairable systems include a system category called consumable, and thus they are non-repairable system at low value. The Repairable systems are the ones that are repaired when they fail, that is when they are not longer operating at 100%, such as a typical example of non-repairable system is the engine. A system is not operative when a component or a subsystem is damaged, then it is substituted or repaired; usually a Repairable system is the one at high economic value where a component substitution/repair cost is more convenient rather than new system purchase. These following parameters: the MTTR “MEAN TIME TO REPAIR” and the MTBF “MEAN TIME BETWEEN FAILURES” are necessary to introduce in order to study and analyse Non Repairable systems. For each kind of component we run some simulations in order to quantify the cost of the customer support for a specific logistic scenario and for a defined time (3 years). In particular for the simulation with this software we assume a fleet size of 2-3-5 or 7 aircraft that have to flight for 300 500 and 700 hours for a years ensuring a fleet availability of 70 80 90 %. The simulation results are showed in table 1. It summarizes the quote of spare parts cost for the customer support for repairable components.
TABLE 1 Costs for repairable components On equal total flight hours (for example 4500 FHtot) it can be highlighted that the Flight Hours affects the customer support costs more than the fleet size. It can be because there are more flight hours to be distributed on fewer aircraft used. In figure 5 is showed an example of Cost Effective curves as output of the simulation software Opus assuming a fleet of 7 aircraft (AC) that have to flight for 300 500 700 flight hours (FH) to show the variation of repairable parts cost with increasing FH.
FIG. 6 Variation of Repairable parts Curve Cost with increasing numbers of aircraft The result is a shift to the right due to an increase in costs as the N°AC. But this time it is less obvious compared to the previous case; it confirms the higher influence of the FH compared to the N°AC on the spare parts costs. The histograms of the simulations results were built (fig 7). It shows the trend of the repairable components cost. These costs are parameterized respect to a given "K factor“ for the privacy of the company. Trend Rip 300FH 2,5 y = 0,2008x + 1,2124
Fattore "K"
2 1,5
A70%
y = 0,1346x + 0,8746
A80%
y = 0,101x + 0,7181
1
A90%
0,5 0 2
3
5
7
n° AC
Trend Rip 500FH 3,5 y = 0,3039x + 1,784
3 2,5 Fattore "K"
FIG. 5 Variation of Repairable Curve Cost with increasing FH This diagram shows that the cost increases as FHs per year, in particular, on equal number of aircraft the curves are shifted to the right, and it led to an increase of the cost with the same trend. Then the figure 6 shows the variation of repairable components cost with increasing numbers of aircraft.
2 1,5
y = 0,2427x + 1,1379
A70%
y = 0,1821x + 0,9554
A80% A90%
1 0,5 0 2
3
5 n° AC
7
aircraft on the spare parts costs.
Trend Rip 700FH 4,5 y = 0,5553x + 1,9186
4
Fattore "K"
3,5 3
y = 0,3404x + 1,4346
2,5
A70% A80% A90%
y = 0,2842x + 1,1126
2 1,5 1 0,5 0 2
3
5
7
n° AC
FIG. 7 Repairable trend for 300 FH, 500 FH, 700 FH The trend can be considered, with good approximation, linear, the curves are almost parallel to each other and there is the same behavior in the case of 300 and 500 and 700FH. The same procedure are used for the definition of costs for non repairable and consumable systems. Then in the tables 2 and 3 are reported the spare parts costs for the non repairable and consumable components.
C. All simulations envelope The goal is to make an envelope of simulations but in order to obtain it the reduction of the number of variables and thus the number of plots on which to operate it is necessary. Now we eliminate in this respect the distinction between repairable materials, consumables and non-repairable, thus obtaining a table of cumulative cost in which each item of cost is the sum of the costs of all 3 types of materials (table 4). N° AC
FH
FHx3years
A 70%
A 80%
A 90%
2
300
1800
1,00
1,23
1,70
3
300
2700
1,09
1,34
1,86
5
300
4500
1,28
1,60
2,19
7
300
6300
1,46
1,80
2,47
2
500
3000
1,41
1,69
2,45
3
500
4500
1,58
1,92
2,75
5
500
7500
1,91
2,35
3,19
7
500
10500
2,25
2,70
3,68
2
700
4200
1,74
2,19
3,00
3
700
6300
2,04
2,52
3,40
5
700
10500
2,56
3,02
4,16
7
700
14700
3,01
3,66
5,11
TABLE 4 Cumulative costs Thus it is possible to buil onlt In questo modo è possibile costruire solo 3 istogrammi diagrammando i costi totali in funzione delle FH. Sorting data following the increasing of total Fh it is possible to incorporate two variables into one, namely the number of aircraft and the FH, allowing the construction of a single histogram showing the evolution of cumulative total costs at total FH and request Availability variations (fig. 9). TABLE 2 Costs for non repairable components
TABLE 3 Costs for consumable components Then the C/E curve is built also for the non repairable and consumable components; also these cases confirm the higher influence of the flight hours compared to the number of
better interpolate all known points found with an error less than15%. The goal is to define an algorithm that allows us to estimate the cost of the customer support for a specific fleet.
Trend costi cumulati a 300FH 3,00 2,50
y = 0,2635x + 1,3974
k factor
2,00 1,50
y = 0,1967x + 0,9991
A70%
y = 0,1587x + 0,8125
A80% A90%
1,00 0,50 0,00 2
3
5
7
n° A/C
Trend costi cumulati a 500FH
IV. THE DEFINITION OF THE ALGORITHM Per definire l’algoritmo sono stati diagrammati tutti i costi cumulati in funzione delle FHtot ottenendo tre curve, ognuna delle quali relativa al livello di disponibilità richiesta. The mathematical law analyzed are: the linear low (fig 10), the power law (fig. 11), the 5th degree polynomial law (fig. 12), the exponential law (fig. 13), the logarithmic law (fig. 14).
4,00 y = 0,4113x + 1,9907
3,50
k factor
3,00 y = 0,3459x + 1,2998
2,50
A70%
y = 0,2835x + 1,0766
2,00
A80% A90%
1,50 1,00 0,50 0,00 2
3
5
7
N° A/C
Trend costi cumulati 700FH 6,00 5,00 4,00 k factor
FIG. 10 Cumulative costs - Linear law
y = 0,7108x + 2,1399
3,00
y = 0,4913x + 1,6153
A70%
y = 0,4309x + 1,2599
A80%
Cumulativecosts–powerlaw
A90% 2,00 1,00 0,00 2
3
5
7
n° A/C
FIG. 8 Cumulative costs trend for 300 FH, 500 FH, 700 FH.
FIG. 11Cumulative costs – Power law Cumulativecosts5thdegreepolynomiallaw
FIG. 12 Cumulative costs – 5th degree polynomial law FIG. 9 Cumulative costs trend. All available data are collected and plotted in one diagram and the next step is to define the mathematical law which
Cumulativecosts–exponentiallaw
FIG. 13 Cumulative costs – exponential law Cumulativecosts–logarithmiclaw
TABLE 6 The algorithm output with power law
TABLE 7 The algorithm output with 5th degree polynomial law The linear law and the power law provide value that are within the error range of 15% assumed before. Then we have done other tests. The results is that the linear algorithm provides values that are outside the error range we have assumed before and so only the power law returns good output. V. The algorithm testing Another test is the one using real data provided from past business negotiations. In tables 8 and 9 are reported two different example of real past negotiations.
FIG. 14 Cumulative costs – logarithmic law TABLE 8 Features of a business negotiation 1
From a first graphical analysis, analyzing the acceptable error range, the mathematical laws that interpolate better the previous diagram are the power law, the linear law and the 5th degree polynomial law. A. The algorithm output Then we can test these laws with the results of the accurate estimation provides by simulation software. Let’s consider this situation: a fleet of 3 AC that have to flight for 300 hours and a customer support period of 3 years. In this situation we have 2700 total FH. The total FH are used as input in mathematical laws and the results are showed in tables 5, 6 and 7.
TABLE 9 Features of a business negotiation 2 Also in these cases the power law algorithm provides good results (tables 10 and 11).
TABLE 10 The result for the business negotiation 1
TABLE 11 The result for the business negotiation 2 VI. Conclusion
TABLE 5 The algorithm output with linear law
It is possible to conclude that in a RFQ circle, when the uncertainty to succeed in a commercial negotiations, it is possible to avoid the method of an accurate estimation, for sure very precise but extremely expensive in terms of time spent and therefore company cost. By using this tool in Excel of simple use, it is possible to obtain a quick feedback on spare parts cost estimation with good accuracy (around 15%).
This margin is quite acceptable related to the amount of saved time and to the possibility to provide, almost instantly, giving an order of representative magnitude of associated cost to the allocation of spare parts for logistical support in terms of assigned operational scenario. REFERENCES [1]
[2]
[3]
[4]
[5] [6] [7]
S. Castagne, R. Curran, A. Rothwell, M. Price, E. Benard and S. Raghunathan, “A generic tool for cost estimating in aircraft design”, Research in Engineering Design, Volume 18, Number 4, 149-162, 2008. Grisi R.M., Murino T., Zoppoli P., “Risk in Supply Networks: The Case of Aeronautical Firms”, Business Performance Measurement and Management,Springer, 2010 Jayashankar, Smith, Sadeh, “A Multi Agent Framework for Modeling Supply Chain Dynamics”, Technical Report, The Robotics Institute, Carnegie Mellon University, 1996. Lee, Billington. The Evolution of Supply Chain Management Models and Practice at Hewlett-Packard. Interfaces 25 (pp. 42-63). SeptemberOctober, 1995. Marx, W.J., Mavris, D.N., and Schrage, D.P., “A Hierarchical Aircraft Life Cycle Cost Analysis Model” AIAA Paper 95-3861, Sept. 1995. Yassine, Ali A., Parametric design adaptation for competitive products. Journal of Intelligent Manufacturing, 2010 W Wei, M Hansen, Cost economics of aircraft size, Journal of transport Economics and Policy, 2003.
T. Murino graduated in Mechanical Engineering, is assistant professor in the ING-IND 17, Industrial Plant System disciplinary group, in the Faculty of Engineering at University of Naples Federico II. She teaches Manufacturing System Management, Goods and Services Production System, and Industrial Logistics at Engineering Faculty. She is also Professor at “Consorzio Nettuno”. She is also peer-reviewer for Elsevier Editorials, and other journal ISI indexed. The research activities is mainly concerned about the following topics: Simulation modelling; Maintenance strategies; Supply Chain Management models; Quick Response Manufacturing; Sustainable production processes; Location-Routing and vehicle routing Problem, Lean Service and Lean production implementation. Her email address is
[email protected]
G. Naviglio is a PhD student of “Production Systems and Technologies” at the Department of Materials Engineering and Operations Management of the University of Naples “Federico II”. He has a master degree in management engineering from the University of Naples “Federico II”. His research interest is on modeling and simulation of production systems and on Lean Manufacturing . His e-mail address is
[email protected]. E. Romano
