Operations and support cost reduction
an aerospace project management challenge
Special Topics – Aerospace Industry
Jeffrey L. Riggs, University of Alabama-Huntsville, Huntsville, Alabama
The realities of increasing pressure to reduce the national deficit and the accompanying reductions in defense spending are well documented by Saunders  and Morrocco . The era of Gramm-Rudman and Desert Storm have served to underscore a significant trend for the future of military technician training and weapon system availability. There will apparently be continuing pressure to reduce operational and support costs in general, and the length of training courses in particular. But, them is also a definite need to offset reductions in military readiness (or system availability) which may resuit from the necessarily long logistic delays associated with potentially far-flung deployments. The question that naturally arises is “how may the length of a course be reduced while minimizing the impact on system availability?”
Several related problems have been addressed. They include:
- Prediction of repair times for the purpose of influencing the design process ;
- Identification of maintenance policy alternatives which reduce failure frequency and/ or failure severity ;
- Development of mathematical models as an aid to military manpower planners with the intent of determining the number of personnel and their skills that best fit the future operational requirements of an enterprise ; and
- Application of probability concepts to repair policies .
The relationship between “repair crew” training, the time required to perform the repair, and the impact on system readiness and operation and support costs has not been quantified.
The purpose of this paper is to present an overview of a method which addresses the relationship between system readiness and technician training. This naturally leads to a broader discussion of operations and support cost reduction opportunities and the broad challenge they present to aerospace project managers.
To properly address the intricacies of operation and support cost, a distinction must be drawn between effective and efficient courses of training. Effective training is defined as any level of training such that operational requirements are met. Efficient training is defined as the minimum amount of training such that minimum operational requirements are met. It would seem that an era is passing during which effective training was a satisfactory goal. The new era will have as its goal efficient training. Again, the question arises “How may the length of a course be reduced while minimizing the impact on system availability?”
A quantitative relationship between availability and technician training requires an expanded understanding of the individual lessons which make up a course, knowledge of the implications of lesson reduction on maintenance task completion time, and reliability data that give insight into the frequency at which maintenance tasks are performed.
NATURE OF THE PROBLEM
Each subsystem has its associated series of maintenance tasks. Each of these tasks has a rate of occurrence. If a task is corrective maintenance, then its rate of occurrence will be governed by the subsystem's reliability. If a task is preventive maintenance, then its rate of occurrence is defined by maintenance policy In addition, each task will have its respective current completion time which can be obtained from field data. A presumption of the approach presented here is that associated with each completion time is a maximum task completion time and a minimum task completion time. Also associated with each task is a set of lessons whose completion and mastery will presumably result in an average completion time being equal to the minimum task completion time. Likewise, in the other extreme, if the contents of the lessons are not grasped by the student, the average completion time will equal the maximum task completion time. These associations are assumed to be linear across the respective ranges. The impact of the individual Military Occupational Specialty (MOS) course lessons on the task are elicited from subject experts as weighted percentages. The sum of the percentages for the respective lessons associated with a given task must be 100 percent.
Each lesson within the course has its associated length, designated here as POI (Program of Instruction). In a fashion similar to the assignment of maximum and minimum task completion times, each lesson is assigned a minimum feasible time (that time below which there is no value in having a lesson) and maximum feasible time (that time above which there is no measurable reduction in task completion time). As a starting point, each lesson is assigned the maximum time deemed useful by the subject experts. The methodology is dependent on a running tabulation of the amount of time which is left to be taken from each lesson and the amount of time by which each task's average completion time can be increased. As maintenance time increases, system readiness decreases due to the system being down for maintenance. However, if maintenance time is decreased through better training for technicians, then the cost of training will increase. But, it is possibleto imagine a situation where the technician training had increased to such a point as to reduce the total requirements for fulltime maintenance personnel, thus offsetting increased training costs.
The MOS training course used as a “straw man” in the development of this methodology was HAWK Repairer Course, POI MOS 104-27K10 . This course consists of 149 lessons with a current total POI length of 1151 hours. There are 40 different maintenance tasks supported by this course.
Data collection for the lessons began with identifying the four HAWK Missile System subsystems that are linked to the POI for the MOS 104-27K10 course. These include Continuous Wave Acquisition Radar (CWAR), High-Powered Illuminator Radar (HIPIR), Platoon Command Post (PCP), and Tracking Adjunct System (TAS).
The POI also defines the linkages between the individual lessons and their corresponding subsystem tasks. Instructors from the U.S. Army Ordnance Missile and Munitions Center and School (OMMCS) HAWK Division were provided a listing of all the lessons and current POI times. They were asked to provide minimum feasible and maximum feasible lesson times. Each lesson was assigned a minimum feasible time (that time below which there is no value in having a lesson) and maximum feasible time (that time above which there is no measurable reduction in task completion times).
The key issue is how to identify a course reduction which will result in the least increase in mean active maintenance time.
The four subsystems supported by, and the 40 tasks performed by, MOS 104-27K10 were analyzed based on data from the U.S. Army Missile Command (MICOM) Sample Data Collection Program, which is a centralized repository of reliability and maintainability data collected by U.S. Army units all over the world. Instructors in the OMMCS Hawk Division correlated information provided by personnel responsible for the Sample Data Collection database. Once the data was correlated to the task level, the number of failures and the time to repair each occurrence was obtained. Using the appropriate total operating time of the subsystem, the rate of occurrence was calculated. Also derived from these reports were the high, current, and low task completion times. The high times were replaced with maintenance personnel estimates to explore the sensitivity of the methodology.
Achieved availability (Aa) was chosen as the measure of system readiness rather than the more familiar operational availability (Ao). Mathematical definitions of these terms are presented by Blanchard  and Shepelavey :
Ao is the probability that a system, when used under stated conditions in an actual operational environment, will operate satisfactorily.
Ao is dependent on administrative and logistic delays which are not affected by MOS technical training. Therefore, Ao was excluded as a candidate for the methodology figure of merit. The alternative figure of merit is defined as A..
Aa is the probability that a system, when used under stated conditions in an ideal support environment (readily available tools, spares, personnel, etc.) will operate satisfactorily.
Since Aa is dependent on mean active maintenance times which are directly related to MOS technical training, Aa was chosen as the methodolgy, Aa was merit. This figure of merit can be used to compare the relative efficiencies of candidate POI lesson mixes. Alternatively a search can be performed for that POI lesson mix which provides a maximum&
Figure 1. Actual vs. Methodology Aa
The basic premise for relating a POI mix to its respective Aa is the relationship between a lesson's time, the completion times of the tasks which it supports, and the task's rate of occurrence.
Fortunately, common sense and mathematics lead us to the same conclusion concerning optimal course reduction. An optimal course reduction gives the best availability for the specified course reduction.
The key issue is how to identify a course reduction which will result in the least increase in mean active maintenance time. The approach is to reduce the course, one lesson at a time, with the lesson reduced being chosen by searching iteratively through all lessons for that single lesson whose reduction increases least mean active maintenance time.
A convenient way to identify the POI lesson length mix now presents itself. For the MOS 104-27K10 course, each analysis begins with course length set at 1180 hours, the sum of all the course lessons' maximum feasible times. The methodology requires iteration through all 149 lessons, once for each incremental decrease in course length. When the desired Aa or course length is achieved, the process ceases.
Figure 1 graphically portrays the dramatic results of this analysis. For each course length, the respective Aa is the maximum possible in terms of all possible mixes of lessons at the specified course length. For the four subsystems maintained by MOS 104-27K10, the actual Aa based on historical data and the current POI lesson mix is approximately 76 percent. On the other hand, if the course were structured in an optimal POI lesson mix, then an Aa of approximately 96 percent could be achieved with no further investment of resources! Alternately, if a 76 percent Aa is acceptable, then this can be achieved while reducing course length to approximately 930 hours. The course reduction will carry with it an overall reduction in training costs.
Perhaps a more attractive alternative would be to reduce course length (approximately 975 hours), increase availability, and dramatically reduce operations and support costs by reducing the number of full-time maintenance personnel.
The methodology presented here is believed to be a new and sound means of either analyzing the impact of course reductions on availability or exploring the possibility of restructuring an existing course's content while achieving optimal Aa. As with any model, the validity of the model is dependent on the source data.
However, once the database for a given course has been constructed and validated by appropriate maintenance (tasks) and training (lessons) personnel, the model can be used with the normal common sense checks. Future areas of potentially fruitful work include:
- Developing task completion time data collection programs which parallel those in existence for collecting handware failure data.
- Analyzing task occurrence data to determine which tasks are encountered so infrequently as to require refresher courses.
- Targeting those tasks with high completion time as candidates for technician expert system support.
- Identifying those lessons which can be completed after assignment to an operational unit instead of while the student is still in the formal training course.
- Incorporating the Delphi technique, analytic hierarchy process, and non-parametric statistical techniques into the process of collecting subjective data.
- Identifying hardware/software initiatives which can enhance the effectiveness of maintenance operations and training.
This type of analysis should be a cornerstone in upcoming operations and support cost reduction efforts.
1. Saunders, Norman C. 1990. Defense Spending in the 1990s-The effect of deeper cuts. Monthly Labor Review, vol. 113, (October) pp. 3-15.
2. Morrocco, John D. 1990. Cheney Proposes Stretchouts, Cuts in B-2, C-17 Programs. Aviation Week and Space Technology, vol. 132, (April) pp. 18-20.
3. Wohl, Joseph G. 1982. Maintainability Prediction Revisited: Diagnostic Behavior, System Complexity, and Repair Time. IEEE Transactions on Systems, Man, and Cybernetics, vol. 12, no. 3, (June) pp. 241-250.
4. Hardy, S.T., and Krajewski, L.J. 1975. A Simulation of Interactive Maintenance Decisions. Decision Sciences, vol. 6, (January) pp. 92-105.
5. Gass, Saul I. 1991. Military Manpower Planning Models, International Journal of Computers & Operations Research, vol. 18, no. 1, pp. 65-73.
6. Shephard, R.W. Hartley, D.A., Haysman, P.J., Thorpe, L., and Bathe, M.R. 1988. Applied Operations Research. New York, NY: Plenum Press. pp. 103-105.
7. U.S. Army Ordnance Missile and Munitions Center and School. 1989. Program of Instruction for 104-27K10 HAWK Repairer Course. Redstone Arsenal, Alabama: U.S. Army.
8. Blanchard, Benjamin S. 1986. Logistics Engineering and Management.. Englewood Cliffs, NJ: Prentice-Hall, pp. 32-47.
9. Shepelavey, Bohdan. 1990. Operational Availability and Its Maximization. Logistics Spectrum, vol. 25, p. 257.
Jeffrey L. Riggs is on the faculty at the University of Alabama in Huntsville, teaching courses in systems engineering, operations research, and management science. He received his B.S. degree from the U.S. Naval Academy, M.S. degree in industrial engineering from the University of Missouri, M.A. in management from Webster University, and Ph.D. in systems engineering from the University of Alabama in Huntsville. His previous experience includes six years in the Marine Corps as a radar maintenance platoon leader, HAWK missile firing battery commander, R&D project engineer, and 12 years as a civilian operations research analyst and project engineer.
AUGUST 1992 pm network