Reserve Demand Levels

Abstract

A crew planning system includes a demand forecasting module and an optimization module. The system forecasts anticipated reserve demand levels, and determines suitable reserve staffing approaches to meet anticipated reserve demand. The forecasting is based upon probabilistic distribution models which take into account variability associated with reserve demand for a particular day. Via use of the crew planning system, reserve staffing expenses may be reduced and/or reserve demand may be met with a higher degree of probability.

What is claimed is: 
     
         1 . A method comprising:
 forecasting, by a computer, a first daily reserve demand level for each day in a first time period,   wherein the first daily reserve demand level for each day is generated based on a mean and a standard deviation of historical reserve demand level for that day, a minimum probability that a selected reserve pattern will exceed the first daily reserve demand level, and a mapping coefficient configured to relate the reserve pattern to a corresponding day in the first time period;   satisfying, by the computer using a risk-based solution, the first daily reserve demand level for each day in the first time period; and   utilizing, by the computer, the performance of the risk-based solution during the first time period to create a second daily reserve demand level for each day in a second time period.   
     
     
         2 . The method of  claim 1 , wherein the forecasting comprises:
 preparing, by the computer, a subset of historical open duty periods (ODPs) and block time data based on specified filters;   assigning, by the computer, a plurality of indicator variables to each day in the first time period, wherein at least two of the plurality of indicator variables are configured with a dominant relationship;   fitting, by the computer, a linear regression model of the historical ODP as a function of the block time data and the plurality of indicator variables, wherein the results of the linear regression model comprise coefficients for the block time data and for each indicator variable; and   utilizing, by the computer, the linear regression model to estimate the mean and standard deviation of each of the historical ODPs.   
     
     
         3 . The method of  claim 1 , further comprising evaluating, by the computer, the performance of the risk-based solution against an actual daily reserve demand level for each day in the first time period. 
     
     
         4 . The method of  claim 1 , wherein the forecasting utilizes a cumulative distribution function and based on a minimum number of reserve crew to cover the first daily reserve demand level, a minimum probability that a reserve pattern will exceed the first daily reserve demand level, a first risk variable representing a minimal acceptable risk of not covering the first daily reserve demand level, a financial cost of the reserve pattern, and a cancellation cost of cancelling one or more flights in the first time period. 
     
     
         5 . The method of  claim 1 , further comprising:
 generating, by the computer using a probability distribution reserve staffing model, a lower bound of a reserve airline staffing level for each day in the first time period based on the first risk variable representing a minimal acceptable risk;   modifying, by the computer, the reserve airline staffing level for a first day in the first time period based on a second risk variable that includes updated risk information;   modifying, by the computer, the second risk variable associated with the reserve airline staffing level for a second day in the first time period;   modifying, by the computer, the reserve airline staffing level for the second day in the first time period based on the second risk variable that includes updated risk information; and   allocating, by the computer, reserve staff based on the reserve airline staffing level.   
     
     
         6 . The method of  claim 5 , further comprising:
 evaluating, by the computer, a result of the probability distribution reserve staffing model for each day in the first time period to determine reserve airline staffing results information; and   utilizing, by the computer, the reserve airline staffing results information to forecast the the second daily reserve demand level in a second time period.   
     
     
         7 . The method of  claim 5 , wherein the forecasting further comprises at least one of:
 accessing, by the computer, historical reserve demand information, wherein the historical reserve demand information is associated with at least one indicator value;   determining, by the computer, a regression model for the historical reserve demand information utilizing the at least one indicator value;   determining, by the computer, a mean associated with the historical reserve demand information; or   determining, by the computer, a standard deviation associated with the historical reserve demand information.   
     
     
         8 . The method of  claim 5 , wherein the generating using the probability distribution reserve staffing model to generate a reserve staffing level operates based on at least one of:
 each reserve staff member can work for at most 18 days in the first time period;   each reserve staff member must have at least 4 consecutive days off in the first time period; or   each reserve staff member can be assigned to at most 6 consecutive working days.   
     
     
         9 . The method of  claim 5 , wherein the generating further comprises at least one of:
 evaluating, by the computer, a risk of flight cancellation arising from the first daily reserve demand level exceeding a reserve airline staffing level; or   evaluating, by the computer, a financial impact of a flight cancellation arising from the first daily reserve demand level exceeding a reserve level.   
     
     
         10 . The method of  claim 7 , wherein the determining the regression model utilizes a plurality of indicator variables, and wherein the plurality of indicator variables are configured with a dominant relationship. 
     
     
         11 . The method of  claim 10 , wherein the plurality of indicator variables comprises a peak day, a transition day and a weekend, wherein the peak day is dominant over the transition day and the weekend, and wherein the transition day is dominant over the weekend. 
     
     
         12 . The method of  claim 5 , further comprising reducing, by the computer, a number of reserve staff for each day in the first time period, based at least in part on the reserve airline staffing level. 
     
     
         13 . The method of  claim 5 , further comprising varying, by the computer, the first risk variable associated with the first day in the first time period, wherein the first risk variable represents a likelihood of the reserve airline staffing level exceeding the first daily reserve demand level on the first day. 
     
     
         14 . The method of  claim 5 , wherein the probability distribution reserve staffing model is formulated as: 
       
         
           
             
               
                 min 
                 ⁢ 
                     
                 Cost 
               
               = 
               
                 
                   ∑ 
                   
                     j 
                        
                     ∈ 
                     P 
                   
                 
                 
                   x 
                   j 
                 
               
             
           
         
         
           
             
               subject 
               ⁢ 
                   
               to 
             
           
         
         
           
             
               
                 P 
                 ⁡ 
                 ( 
                 
                   
                     
                       ∑ 
                       
                         j 
                            
                         ∈ 
                         P 
                       
                     
                     
                       
                         A 
                         
                           j 
                           ⁢ 
                           t 
                         
                       
                       ⁢ 
                       
                         x 
                         j 
                       
                     
                   
                   ≥ 
                   
                     d 
                     t 
                   
                 
                 ) 
               
               ≥ 
               
                 
                   α 
                   t 
                 
                 ⁢ 
                      
                 
                   ∀ 
                   
                     t 
                     ∈ 
                     T 
                   
                 
               
             
           
         
         
           
             and 
           
         
         
           
             
               
                 x 
                 j 
               
               = 
               
                 { 
                 
                   
                     
                       
                         
                           1 
                           ⁢ 
                               
                           if 
                           ⁢ 
                               
                           pattern 
                           ⁢ 
                               
                           j 
                           ⁢ 
                               
                           is 
                           ⁢ 
                                
                           included 
                           ⁢ 
                                
                           in 
                           ⁢ 
                                
                           the 
                           ⁢ 
                               
                           solution 
                         
                       
                     
                     
                       
                         
                           0 
                           ⁢ 
                               
                           otherwise 
                         
                       
                     
                   
                   ⁢ 
                       
                   
                     ∀ 
                     
                       j 
                       ∈ 
                       P 
                     
                   
                 
               
             
           
         
         where: 
         ∝ t  represents the minimum probability that the reserve pattern will exceed the first daily reserve demand level, 
         C j  represents the financial cost of selecting the reserve pattern j as part of the selected solution, 
         C′ t  represents the financial cost of having insufficient reserve crews to cover a scheduled set of airline flights in the first time period, 
         t represents a day in the first time period, 
         x j  represents a binary decision variable to include reserve pattern j as part of the selected solution, and 
         A jt  represents a mapping coefficient configured to relate the reserve pattern j to the corresponding day tin the first time period. 
       
     
     
         15 . The method of  claim 5 , wherein the generating comprises:
 determining, by the computer, an unsatisfied demand cost associated with the first daily reserve demand level exceeding the reserve level on the first day in the first time period;   determining, by the computer, an excess airline staffing cost associated with the reserve airline staffing level exceeding the first daily reserve demand level on the first day; and   weighting, by the computer, the unsatisfied demand cost and the excess airline staffing cost to determine whether to modify the first daily reserve level on the first day.   
     
     
         16 . The method of  claim 5 , wherein:
 the reserve airline staffing level comprises a set of reserve members,   the first day is a peak day,   the second day is a non-peak day, and   the modifying the reserve airline staffing level for the first day and the varying the second risk variable associated with the second day are accomplished with a fixed number of reserve members.   
     
     
         17 . The method of  claim 5 , further comprising implementing, by the computer, the reserve airline staffing level by staffing an airline flight with a reserve staff member in place of a regularly scheduled member. 
     
     
         18 . The method of  claim 5 , wherein the tuning includes placing frequently used files on separate file systems to reduce in and out bottlenecks. 
     
     
         19 . The method of  claim 5 , wherein the probability distribution reserve staffing model is further based on a financial cost of selecting a reserve pattern as part of a selected solution, and a financial cost of having insufficient reserve crews to cover a scheduled set of airline flights in the first time period. 
     
     
         20 . The method of  claim 1 , further comprising:
 storing, by the computer, the reserve airline staffing level in a database;   tuning, by the computer, the database to optimize database performance;   designating, by the computer, the reserve airline staffing level as a key field in a plurality of related data tables to speed searching for the data;   linking, by the computer, the plurality of related data tables based on the type of the reserve airline staffing level in the key fields; and   obtaining, by the computer, the reserve airline staffing level from the database.