An Integrated Approach to Indirect Workforce Planning for Semiconductor Wafer Fabrication Operation

Sih Wei Sing

Abstract

Manpower planning model is gradually becoming an essential tool for the managers of semiconductor company to gain insight of the capability of the workforce. With the knowledge of possible demand variation in the future, managers can strategically plan the workforce to achieve their targeted goals; increase manufacturing productivity and maintain cost-effectiveness. Many semiconductor companies plan and forecast the manpower requirements based on group expert decision-making. It is a judgmental technique that draws from varying degrees of past experiences, opinions, ideas and motivation from a pool of experts/managerial engineers and then decides on the manpower allocation. However, the accuracy of the forecast yielded through this technique is doubtful as the forecast is heavily dependent on the opinions of the managers which could be plagued by human biases.

The manpower of the fabrication is divided into two categories, namely the direct workforce and indirect workforce. This paper will focus on the indirect workforce and propose an approach which integrates a series of models to forecast and optimize the indirect workforce requirements based on the targeted fabrications outputs. There are three phases to this approach. In Phase 1, the targeted outputs forecasts are adjusted using an Analytic Hierarchy Process (AHP) based method. Pair-wise comparisons of the contributing criteria and manpower are performed to obtain their weights of importance w.r.t the targeted outputs. The forecasts from each contributing engineering department are adjusted and aggregated proportionally to these weights to obtain a single set of forecast for the targeted output set for each fab.

In Phase 2, Data Envelopment Analysis (DEA) inspired Linear Programming (LP) models are formulated. They allow manger to forecast the optimal/minimum total headcount of indirect workforce for the company based on the historical records of the semiconductor company and the adjusted wafer fabrication output set forecasts from Phase 1. The LP models are solved using the software ILOG OPL-IDE.

In Phase 3, a Goal Programming (GP) model is developed to allocate the indirect workforce to respective fabrications owned by the semiconductor company with minimum impact on cost incurred due to labour shortage or excess. Achieving the optimal headcount to each fab and the entire company (obtained from Phase 2) without upsetting the predetermined workforce turnover rate are the goals of this allocation model. The results generated will help the manager to decide the distribution of indirect workforce across the semiconductor company.

As an example on the application of this integrated approach, the data of an established semiconductor company in Singapore was put through the series of models developed. The integrated approach brought attention to translating subjective managerial judgments into numerical values so that the intangible factors can be incorporated into the forecast and be determined with more mathematical rigour. The results of this application show that the adjusted forecasts perform better in both Mean Absolute Deviation (MAD) and Mean Squared Error (MSE). Based on the company data, the cost incurred by allocating workforce using the integrated approach is also found to be lower than the current practice.

A comparison between the integrated approach adopted in this paper and the indirect workforce allocation model Chien et al. (2008) is made. The integrated approach is preferable in 2 areas; monetary and specialization of workforce. The approach proposed is also administered theoretically and discussed under two possible scenarios; surge in wafer demand and changes in turnover rates.

The proposed approach is applicable to many other companies that have the similar structure. However minor modifications have to be made, generalizing the model and analysis to accommodate some of the other elements was deemed beyond the scope of this paper.