Patent Application
20060111871
1. Field of the Invention
The present invention relates generally to maintenance demand planning and optimization, and more particularly to a method of and system for generating a service plan for a group of one or more assemblies that is responsive to information regarding unscheduled events such as unscheduled service events relevant to the assemblies.
2. Description of the Background of the Invention
In a group of assemblies wherein each assembly includes a number of parts that deteriorate with use, each assembly will require periodic maintenance servicing to replace deteriorated parts. Generally, an owner of the group of assemblies desires to minimize the total cost of the maintenance for the assemblies for the duration of a given time period such as, for example, during the contract period with a maintenance organization. The maintenance cost can be reduced by developing a maintenance plan based on a schedule for the periodic maintenance in order to replace the deteriorated parts before they become unusable, provided that the parts wear according to a predictable schedule. Parts with predictable wear patterns have a known limited useful lifespan and are referred to as life limited parts (LLP's). In addition, the assemblies are often mounted to individual supporting structures such that the maintenance for each assembly is performed with the assembly either mounted to the associated supporting structure or removed from such structure. The cost of repair is generally greater when the assembly must be removed from the associated supporting structure than when the assembly can remain mounted to such structure. This removal cost increase is a result of the extra labor required to remove and replace the assembly, the extra time that the overall system in which the assembly is used is out of service (the total costs associated with the time that the system is out of service is referred to as downtime costs), and the extra maintenance procedures required by removal of the assembly from the structure (e.g., repairs performed “at next removal”). These removal costs militate in favor of complete replacement of worn parts when an assembly is removed from the structure in order to reduce the need for future assembly removals.
A competing cost consideration, however, relates to the cost incurred when LLP's are replaced, both in terms of the replacement cost of the part and the cost associated with any wasted useful lifespan of the LLP. These replacement costs militate in favor of more frequent servicing so that the full usable life of each LLP can be used. This plan, however, if implemented, can undesirably increase overall costs to the operator of the system due to increased labor and downtime costs, particularly when the assembly must be removed from the supporting structure. Further complicating maintenance considerations is the fact that the maintenance organization desires to operate in as efficient a manner as possible to keep costs to a minimum. Specifically, the maintenance organization usually wants to plan for servicing events so that parts are available on a just-in-time basis and so that servicing personnel are utilized to as close as possible to full capacity without becoming a bottleneck in the servicing process.
From the foregoing, it becomes apparent that there is a trade off between minimizing the costs of removing the assemblies from the support structure to access the assembly and minimizing the costs associated with the unused life of the LLP'S. At one extreme, each LLP could be used for its full life span. To accomplish this, the assembly may need to be removed from the associated structure every time an LLP reaches the end of the usable lifespan thereof. This would minimize wasted useful lifespan of the LLP's, but it would also greatly increase the removal costs for the assembly during the lifespan thereof. At the other extreme, all the LLP's in the assembly could be replaced every time the assembly is removed from the structure. This would minimize the removal costs for the assembly by reducing the frequency of removing the assembly from the support structure, but it would greatly increase the wasted useful lifespan of the LLP's. The optimal solution, however, is usually somewhere between these two extremes. In essence, a balance must be struck between more frequent servicing so that LLP's can be used closer to the end of the useful lives thereof, and less frequent servicing so that the removal costs can be controlled. By so balancing, an optimal service plan may be developed in which the overall service costs are minimized while keeping operator disruptions and servicing organization inefficiencies to a minimum.
Ideally, all maintenance for replacing LLP's in an assembly is planned according to a schedule of maintenance events based on the known remaining lifespan of the individual LLP's. Similarly, maintenance for each assembly in the entire group of assemblies is ideally planned according to an overall maintenance schedule for the entire group. Realistically, however, unplanned exigencies occur that require maintenance to be performed on assemblies that was not scheduled. Such unscheduled maintenance events, similar to the scheduled maintenance events, may require the assembly to be removed from the associated supporting structure. A problem with this is that an unscheduled maintenance event is necessarily more expensive than a scheduled maintenance event of similar scope because of the resulting disruption to the overall maintenance schedule.
An example of such a maintenance organization is one that services and maintains a group of aerospace products such as a number of gas turbine engines for a fleet of airplanes. It is desirable to plan a schedule of service events including maintenance, repair, and overhaul (MRO) events for all the engines in the fleet in an attempt to minimize maintenance costs including queuing costs (Kq) associated with queuing delays, parts stocking costs, and unused life costs for each engine (the total unused life costs for an engine is referred to as Kw). All of the MRO events can be categorized as either a scheduled maintenance event (SME) or an unscheduled maintenance event (UME). SME's can be any scheduled MRO event and can be subdivided into two basic sub-categories, scheduled in-place maintenance (e.g., for an automobile, an oil change with no engine removal) and scheduled engine removal (SER) (e.g., for an automobile, an engine overhaul requiring engine removal). Similarly, UME's may also involve a wide range of service levels, including an unscheduled engine removal (UER). UER's are generally caused by relatively random, unavoidable events (from the perspective of the maintenance organization), such as an unpredicted part failure or bird strikes, and can significantly disrupt the optimized workflow and parts stocking that is planned for SME's.
The maintenance organization can develop a maintenance schedule for the SME's that optimizes the overall costs of the fleet maintenance by balancing several potentially competing considerations to obtain an optimal service scope specification—or build to level (BTL)—for every SME. The optimal BTL balances the cost of replacing LLP's before the end of the usable life thereof—which wastes useable life of the part and increases Kw—with the cost of more frequent SER's to obtain a minimized overall servicing cost for both that engine and the overall total number of engines in the fleet. For example, a high BTL is a more stringent specification that increases Kw, but tends to decrease subsequent engine removal rate and removal costs due to the increased useable life spans of the engine and its parts. Conversely, a low BTL is a less stringent specification that reduces the immediate parts replacement cost and Kw by using more of the full usable life span of the LLP's, but tends to increase subsequent engine removal rate and removal costs due to the decreased remaining useable life spans of the engine and its parts. Therefore, the service plan for a particular engine optimizes the BTL for a particular SME to balance between higher parts replacement costs associated with a high BTL and higher downtime costs associated with a low BTL when calculating a maintenance schedule. As a result, LLP's may or may not be replaced during any given SER depending on the optimized BTL for that SME.
A known method of demand optimization for generating a service plan for providing maintenance service to a fleet of aerospace vehicles is schematically depicted in
The purpose of the service plan is to help ensure that each engine and each part in each engine is serviced prior to either the RIm or RIp reaching zero, to help the maintenance organization plan for the flow of engines that will be inducted into a shop for servicing, and to reduce the total cost to the owner of maintaining the engines by choosing one or more optimal BTL's for the maintenance events that will minimize the sum of the service costs (e.g., Kp and Kq, etc.) for the entire fleet during the chosen time period. The fleet-wide service plan includes a schedule of service events for all of the engines over the duration of the chosen time period and a work scope specification of tasks to be performed during each scheduled service event. Each work scope specification includes a set of keep/replace decisions for the LLP's in the engine. If no target BTL has been supplied to the DOP, an optimal BTL, which determines what LLP's are kept or replaced, is generated for each engine at each SME such that the cost of all the SME's across the fleet is minimized for the chosen time period. If a target BTL is provided, the DOP will generate a schedule of SME's, with the work scope specification for each SME meeting the target BTL. Of course, the actual keep/replace decisions for individual LLP's may deviate from the plan once an engine has been inducted into the shop and inspected according to the actual level of service deemed necessary by the servicing personnel. However, because a plurality of LLP's in a particular engine (each having a different RIp value at any given time) may need to be replaced during any given SER, most LLP's will likely have at least some useful life left when they are removed, which directly corresponds to increased part replacement costs Kp by increasing the wasted unused part life. By tracking the RIm and RIp's for each engine in a fleet with inputted status data for the engine specific parameters, the DOP can schedule the SER's and compute optimized BTL's to minimize service costs across the entire fleet, thereby optimizing the scheduled maintenance costs.
Known demand optimization programs, however, have not been used particularly effectively to plan for the occurrence of future UER's. According to the method of
According to one aspect of the invention, a method of service planning includes inputting an actual service parameter representing service required by at least one component of an assembly into a maintenance demand optimization system, and inputting a virtual service parameter representing unscheduled service events into the maintenance demand optimization system. The virtual service parameter is derived from data related to unscheduled service events experienced by a number of assemblies relevant to the assembly. The maintenance demand optimization system is operated to develop a service plan for the assembly.
According to another aspect of the invention, a method of scheduling a task to be performed on an assembly having a plurality of parts includes retrieving condition information for at least one of the parts and retrieving unscheduled event information relevant to the assembly. A schedule for performing the task is generated, wherein the schedule is responsive to the condition information and the unscheduled event information.
According to another aspect of the invention, a method of generating a service plan for a group of assemblies, each assembly having a predictable remaining useful life span, includes retrieving condition information relating to the remaining usable life span of the assemblies and retrieving unscheduled event information relevant to the assemblies. A service plan for the group of assemblies is generated wherein the service plan is responsive to the condition information and the unscheduled event information.
According to another aspect of the invention, a method of generating a maintenance work scope specification for a group of at least one assembly includes retrieving a plurality of possible remaining life span intervals for an assembly representative of the group, retrieving a plurality of possible work scope specification levels for the representative assembly, and retrieving unscheduled event information relevant to the representative assembly. A first service plan option responsive to the unscheduled event information is generated for a combination of a first one of the possible specification levels and the possible remaining life span intervals, and a second service plan option responsive to the unscheduled event information is generated for a combination of a second one of the possible specification levels and the possible remaining life span intervals. A cost is calculated for each of the first and second service plan options, and one of the first or second possible specification levels corresponding to one of the first or second service plan options having a preferred cost is selected.
According to another aspect of the invention, a method of generating a service plan for a group of assemblies, each assembly having a predictable remaining useful life span, includes retrieving a plurality of possible remaining life span intervals for an assembly representative of the group, retrieving a plurality of possible work scope specification levels for the representative assembly, and retrieving unscheduled event information relevant to the representative assembly. A first service plan option responsive to the unscheduled event information is generated for a combination of a first one of the possible specification levels and the possible remaining life span intervals, and a second service plan option responsive to the unscheduled event information is generated for a combination of a second one of the possible specification levels and the possible remaining life span intervals. A cost for each of the first and second service plan options is calculated, and one of the first or second possible specification levels corresponding to one of the first or second service plan options having a preferred cost is selected. Actual condition information relating to the remaining usable life span of the assemblies, the unscheduled event information relevant to the assemblies, and the selected one of the specification levels is retrieved. A service plan for the group of assemblies responsive to the condition information, the unscheduled event information, and the selected one of the specifications is generated.
According to another aspect of the invention, a method of representing unscheduled maintenance events in a maintenance demand planning system, wherein the system generates a service plan for at least one assembly that is responsive to inputted condition information relevant to the assembly, includes retrieving unscheduled maintenance events information relevant to the assembly and deriving a virtual service parameter related to the unscheduled maintenance events information. The virtual service parameter is provided to the system for generating the service plan.
According to another aspect of the invention, a computer assisted system for generating a service plan for a group of assemblies includes means for retrieving condition information and unscheduled event information relevant to at least one of the assemblies and means responsive to the condition information and the unscheduled event information for generating the service plan.
According to another aspect of the invention, a computer assisted system for scheduling a task on a product having a plurality of actual parts includes means for retrieving information from a database, wherein the information includes actual part information related to at least one of the actual parts and virtual part information related to a virtual part that is not included in the product. The virtual part information is related to unscheduled event information relevant to the product. The system also includes means responsive to the actual part information and the virtual part information for generating a task schedule.
According to another aspect of the invention, a method of scheduling a task to be performed on a gas turbine engine includes inputting information relevant to the engine into a maintenance demand optimization system and operating the maintenance demand optimization system to develop a service plan for the assembly. The information includes unscheduled engine removal data.
According to another aspect of the invention, a computer-assisted system for scheduling a service task to be performed on a gas turbine engine includes means for retrieving information from a database relevant to the engine and means responsive to the information for generating a task schedule. The information includes unscheduled engine removal data.
Other aspects and advantages of the present invention will become apparent upon consideration of the following detailed description.
Referring now to
The engine independent parameters 16 include known maintenance cost information and a planning time horizon 26. The planning time horizon 26 is the given time period over which the optimized service plan will be calculated. The known maintenance cost information includes the cost to provide a complete margin restoration for an engine (Km) 28, the additional cost of a scheduled engine removal (Kser) 30, the additional cost associated with an unscheduled engine removal (Kuer) 32, and the replacement costs for each LLP (Kp) 34 being considered. An optional engine independent parameter that may be supplied to the DOP is a desired target maintenance specification, or target BTL 36, for each scheduled service event. If the target BTL 36 is not supplied to the DOP 14, the DOP will calculate an optimal BTL for each scheduled service event during the planning time horizon. If a target BTL 36 is supplied to the DOP 14, however, the DOP will use the target BTL for each scheduled service event to plan the schedule of service events. Thus, providing a target BTL 36 to the DOP 14 reduces the number of parameters to be optimized, which reduces the computational resources necessary to obtain the service plan 12.
The engine specific parameters 18 include a current status parameter RIm 40 for each engine and a current status parameter RIp 38 for each LLP on each engine. Current use status data regarding each engine is supplied so that the RIp and RIm values 38, 40 for each engine in the fleet may be calculated and supplied to the DOP 14. The RIp value 38 at a given time for each part is the difference between the total life expectancy of that part and the used life of the part at that time. The total life expectancy of an LLP is an assigned value representing the acceptable amount the LLP may be used before it must be replaced and is based on past performance of the same or similar parts and acceptable safety factors. The used life (or use status) of each part is a measure of the amount of wear the part has incurred, such as may be tracked by the number of engine use cycles the part has experienced. The RIp value 38 for a given LLP decreases linearly with use until it is replaced. For example, if the total life expectancy of a part is 7 time buckets and the part has been used for 4 time buckets, the RIp value 38 for the part is 3 time buckets. (A time bucket may be any assigned time period, but is generally based on time intervals already being used by the maintenance organization for tracking use data, such as flight cycles.) When the RIp value 38 for a given LLP reaches zero, the part has no useable life left and needs to be replaced. In a similar manner, the RIm value 40 for each engine is the difference between the total life expectancy of the engine before it needs a complete servicing and the current use status of the engine. The current use status of an engine may be measured by any acceptable method known in the aerospace vehicle maintenance industry, such as, for example, by measuring the exhaust gas temperature or counting the number of flight cycles. The RIm value 40 for an engine decreases proportionally with use, and when the RIm value reaches zero, the engine needs to be completely serviced regardless of the RIp values 38 of any of the LLP's in the engine.
The virtual service parameter 20 is derived from statistical data regarding past unscheduled service events related to the engines in the fleet. In this example, the virtual service parameter 20 is derived from UER rate data 42 relevant to the engines in the fleet, but it could be related to the statistical probability of other non-periodic or apparently random maintenance events. The UER rate data 42 is information relating to the statistical frequency of UER's for a sample population of engines relevant to the engines in the fleet being serviced. The virtual service parameter 20 is represented by a virtual part, which does not actually exist on the engines, having a remaining interval value (RIv) 44. The RIv value 44 is supplied to the DOP 14 and processed in the same manner as the RIp and RIm values 38, 40. The RIv value 44 is developed such that as the UER rate for the representative group of engines increases, the RIv value decreases, and vice versa. For example, a higher UER rate for a given population of engines would be represented by a virtual part having a shorter mean life expectancy with a small variance, and is represented by a small RIv value 44. Conversely, a lower UER rate for a given population of engines would be represented by a virtual part having a longer mean life expectancy and a larger variance, and is represented by a larger RIv value 44. The RIv values 44 are typically expected values, but any combination of means and variances can be used, including arbitrary distributions. Although the RIv value 44 may change as the statistical data for UER's changes over time, in general, the RIv value 44 for generating a service plan 12 for a given number of engines is a constant that applies to every engine represented by that RIv value.
In a preferred method of establishing the RIv value 44, historical UER data 42 for the given population of engines is analyzed to determine a statistically probable time interval, number of discrete time buckets, between UER's for those engines. A discrete probability mass function (PMF) based on the probable time interval between UER's is calculated and distributed across the probable number of time buckets. For the following example, a uniform discrete PMF distribution based on the probable time interval is chosen, but other PMF distributions, including non-uniform and continuous distributions, could also be used. An expected value of the mean, EV, is then calculated in which the EV is equal to the sum of the products of each PMF and the time bucket value associated therewith. Next, a truncated EV value, [EV], is derived from the EV value in which the [EV] value is only the integer portion of the EV value. The [EV] value is then used for the RIv value 44 that is supplied to the DOP 14. For example, if UER's are uniformly distributed across time bucket values of {1, 2, 3, 4}, then a PMF of ¼=0.25 representing the probability of a UER for that population is assigned to each of the values {1, 2, 3, 4}; the EV value would be (0.25×1)+(0.25×2)+(0.25×3)+(0.25×4)=2.5; and, the [EV] value would be 2. The RIv value 44 for that population is set equal to the [EV] value, 2. An integer RIv value 44 is used to match the existing format of the RIp and RIm values 38, 40 used in the AMP program, but other formats could be adapted for use with different DOP's and methods. Preferably, the RIv value 44 is formatted and treated exactly the same as the RIp values 38 within the DOP 14 in order to minimize the need for any reconfiguration of the DOP. This method of developing an RIv value 44 is exemplary of just one of many possible methods of developing a virtual service parameter 20 based on past UME occurrences. It is to be understood that the RIv value 44 could also be based on different statistical values developed in other manners such that, as the UER rate for the statistical sample population increases, the RIv value 44 for the virtual part representing the historical UME information would decrease in some relative manner. For example, an RIv value 44 could be based directly inversely on the average UER rate 42.
The DOP 14 retrieves the data regarding each RIp and RIm value 38, 40 for each engine in the fleet, the RIv value 44 relevant to those engines, and the data regarding the engine independent cost parameters 28, 30, 32, 34 and the planning time horizon 26. The DOP 14 also retrieves any data that may be supplied for the target BTL parameter 36. The DOP 14 processes the retrieved data and generates the service plan 12, which includes a schedule 22 of each service event for each engine that is predicted to occur during the planning time horizon 26 and a work scope specification 24 for each scheduled service event. Each work scope specification 24 includes a list of tasks to be performed on the engine being serviced, including individual LLP keep/replace decisions for each LLP on the engine. The purpose of the service plan 12 is to optimize the flow rate of work and minimize the total cost of providing service for the given RIv values and any given BTL value. The DOP 14 accounts for the probability of the occurrence of future UER's when developing the service plan 12 by treating the RIv value 44 in the same manner as the RIp and RIm values 38, 40. The DOP 14 responds to each RIv value 44 in a manner such that if for a given maintenance event it is highly likely that the engine will be back within a short time span for a UER, the service plan 12 may delay a given SER or lower the BTL of a given maintenance event in anticipation of that UER. Conversely, if for a given maintenance event it is highly likely that the engine will not be back in the shop for a long time span for a UER, the DOP may increase the BTL or not delay a given SER for that maintenance event because it is not anticipated that the engine will be back in the shop soon for a UER.
Referring now to
Referring now to
Variations between the actual RIp and RIm values and actual part maintenance requirements in specific engines in the fleet and the projected RIp and RIm values obtained from the optimized service plan 12 will occur as the engines in the fleet are used. When that occurs, it is advantageous to update the service plan 12 to account for the changes in the actual data for the engines to help the maintenance organization adequately prepare for approaching service events. Updated fleet information may become available with every service event, from operation log records, from sensors on the engines such as exhaust gas temperature sensors, or from other sources such as owner guidelines and governmental agency restrictions. To keep the service plan 12 updated, the method 10 is repeated using the optimal target BTL 102 generated with the method 100 as the target BTL 36 supplied to the DOP 14 and providing updated values for the engine specific parameters 18, engine independent parameters 16, and virtual service parameter 20 to obtain an updated service plan 12′. In this manner the service plan 12 may be continually updated to account for changes in the actual fleet status by reiterating the method 10 using the optimal target BTL 118 to account for actual use conditions in the fleet as actual conditions for the engines in the fleet—including the engine specific parameters 18, the virtual service parameter 20, and the engine independent parameters 16—change with time.
It should be apparent from the methods 10 and 100 described previously herein that a service plan that is responsive to historical UER rates relative to a group of engines may be generated and regenerated for different levels of specificity within a fleet, from a single engine to larger subgroups of engines. It should also be apparent that the methods 10, 100 are not limited to being used for servicing engines (since other aircraft components also have LLPs) or even for aerospace vehicle maintenance servicing, but may be easily adapted for use with different applications to generate service plans responsive to unscheduled service event information for groups of other types of assemblies wherein each assembly includes one or more LLP's that require periodic maintenance on a relatively predictable schedule.
Referring now to
A computer program for directing the processing unit 152 to implement the methods 10 and 100 is stored on a computer readable medium 164 and is provided to the processing unit. The program includes the DOP 14 and, preferably, the mean build-to cost calculator program 112 and the mean cost minimizer program 116. In response to the program, the processing unit 152 retrieves the fleet information (26-40 and 44 for method 10; or 26-36, 104, and 106 for method 100) from any of the possible input devices 154, data network 160, or data storage unit 156. The processing unit 152 then processes the retrieved fleet information and generates the service plan 12 according to the method 10 or the optimal target BTL 102 according to the method 100. The service plan 12 and/or optimal target BTL 102 is preferably provided to a user in an intelligible format through any of the output devices 158. The service plan 12 and/or optimal target BTL 102 may also be stored on the data storage unit 156 and transmitted to the other computer systems 162 connected to the data network 160 where the service plan and/or optimal target BTL may be provided to other users on similar output devices in a manner well known in the art.
Numerous modifications to the present invention will be apparent to those skilled in the art in view of the foregoing description. Accordingly, this description is to be construed as illustrative only and is presented for the purpose of enabling those skilled in the art to make and use the invention and to teach the best mode of carrying out same. The exclusive rights to all modifications which come within the scope of the appended claims are reserved.
This invention was made with United States Government support under contract No. 70NANB9H302 awarded by the National Institute of Standards and Technology, under the terms of which the U.S. Government has a paid-up license in this invention and the right in limited circumstances to require the patent owner to license others on reasonable terms.
