Manpower pattern prediction for multi-project situations

Graduation date: 1979The use of logistic-type growth curves in the prediction of future\ud engineering manpower utilization patterns is a viable technique. When a\ud work environment consists of a large number of repetitive type project\ud the use of regression analysis and Weibull curve shapes gives a reliable\ud forecast of the time-phased utilization of manpower. The technique\ud particularly applicable if little is known about the future projects.\ud The technique is also especially relevant when there are a large number\ud of projects which must compete for a variety of resources (various manpower\ud skills in this case).\ud The level at which the most predictable curves were found was dependent\ud upon the organization structure and the project activity network\ud structure. The forecasting procedure is insensitive to both the individual\ud curve shapes and durations when considering the smoothness of the aggregate\ud manpower. However, both were found to be critical factors in the\ud start-up of the forecast.\ud The technique was found to be useful only at an aggregate level,\ud It was not adequate when individual projects or small groups of projects\ud needed to be forecast. The technique is also limited to quarterly or\ud annual forecast periods, Monthly forecasts contained a great deal of\ud irregular random variation.\ud The system has been adopted by the Technical Services (Preliminary\ud Engineering) branch of Oregon Department of Transportation (ODOT) and is\ud now operational there. Computerized input, calculations, manipulation\ud and output are required to make the technique functional. The ODOT\ud system includes a COMPLOT routine that graphically displays the aggregate\ud manpower curves at various levels. Tabular formated output is also\ud used. The system requires considerable technical and managerial input\ud in order to function. This is especially true during the design and\ud implementation stages.\ud Given a prospective list of projects with a tentative timetable for\ud project completion, the system can forecast manpower requirements. A\ud brief description of relevant parameters is required on each project.\ud Such parameters as the number of right-of-way files and the estimated\ud construction cost of structure are applied to regression equations to\ud estimate the manpower required by the various organizational components\ud for each project. This manpower is then time phased by using unique\ud Weibull curve shapes to distribute the total manpower for each of the\ud project's components. This is done in relation to the tentative completion\ud timetable for each project. These time phased estimates for all of\ud the projects are then totaled into a hierarchy of reports. Each report\ud summarizes the manpower needs for different organizations and at different\ud levels. These forecasts can be used for budgeting and planning\ud purposes. The system can also be used in an iterative simulation mode\ud to level manpower. For example, if the initial project array and timetable\ud create an infeasible manpower level, then the array and schedule\ud can be altered through appropriate administrative actions. The revised\ud input can then be resubmitted into the system. This process can be repeated\ud until a feasible (smooth and equal to available resources) manpower\ud pattern is found.\ud This research found that managerial subjective estimates tended\ud to overestimate the manpower needs and underestimate the project duration.\ud An investigation of the relationship between the network of activities\ud making up the projects and the curve shape of the time phased manpower\ud was also done. The number of different types of resources (number of\ud different occupation disciplines in this case) was shown to be the most\ud influential factor.\ud Due to the complex iterative modeling process needed to obtain the\ud regression equations and Weibull curve parameters, the technique will\ud have only limited application in its present form. Additional development\ud of the technique is needed in order to establish simple decision\ud rules that can be applied in general situations