| Summary: | 博士 === 雲林科技大學 === 工業工程博士班 === 99 === In the literature most of resource constrained project scheduling problems (RCPSP) consider the resources which are independent with jobs processing sequence (JPS). But in some industrial manufacture and assembling projects, the resources required are not only independent resources but also dependent resources which depend on JPS. The characteristic of resources constraint is unlikely the RCPSP in the literature. These characteristics of the industry project make particular demands upon the scheduling model and techniques that have to be developed to serve it. But few papers so far have incorporated the dependent resource constrained project scheduling problem.
In this paper we defined a dependent resource constrained project scheduling problem (DRCPSP), and formulated it to two types of model. One is conceptual model, and the other is binary integer programming (BIP) model. Unlike the traditional RCPSP, the DRCPSP models are able to consider both independent and dependent resources which depend on JPS.
The conceptual model was solved by applying Job Sequence Assignment rule (JSA), and the JSA performance was improved by proposed As Early As Possible assignment method (JAEAP).The master problem of BIP model with nonlinear dependent resource constraints was decomposed to solvable subproblems of BIP model with fixed JPS. In order to shorten the CPU time for solving all subproblems, rsBIP algorithm was considered to reduce the number of subproblems and only solve the better subproblems.
The performance of BIP, rsBIP, JSA and JAEAP were tested by sixty-eight combinations of numerical simulation experiments. The experimental results show that BIP algorithm can find the best solution while the number of jobs is smaller or equal to eighteen, rsBIP can solve the number of jobs is smaller or equal to twenty, the JSA and JAEAP algorithms can solve all test problems. The best one of solution quality is BIP, the second is rsBIP, the third is JAEAP, and the last one is JSA.
The DRCPSP was further applied to practical chemical tower construction project. Simulation experiments of eight illustrative problems with 30 runs show that JAEAP outperforms the JSA. Furthermore, from eight groups with a total of 240 problems comparing two common rules adopted in the current industrial practice (CP), JAEAP demonstrates 5.09% ~ 14.60% average reduction in makespan successfully. The simulation results show that JAEAP is more suitable while the available number of construction teams and the range of job durations are in the high level. Therefore, we proposed that small size problems (J <=18) of DRCPSP are solved by method of BIP, middle size problems (18< J <= 20) are solved by method of rsBIP, and large size problems (J >20) are solved by JAEAP
A real-life example is presented to demonstrate the applicability of JAEAP as well, and the sensitivity analysis shows that the number of working teams effects on the variation of minimum makespan. The results can aid project manager consider how to arrange manpower more efficiently.
|
|---|
