The Impact of Artificial Neural networks on Lead Time Reduction in Procurement Processes: case of Petrozim Line Pvt Ltd

Author

NYATHI MICHAEL P

Title

The Impact of Artificial Neural networks on Lead Time Reduction in Procurement Processes: case of Petrozim Line Pvt Ltd

Abstract

The study investigated the impact of ANN on lead time reduction in procurement processes at Petrozim Line Pvt Ltd. Petrozim Line Pvt Ltd, a critical player in Zimbabwe's fuel supply chain, faces significant challenges with extended procurement lead times. These lengthy lead times span the entire procurement cycle, from needs identification through requisitioning, supplier sourcing, negotiation, ordering, and final receipt of goods. On average, it takes Petrozim Line over 90 days to complete a procurement process, leading to operational inefficiencies, stockouts of crucial supplies, increased downtime for essential equipment, and potentially dissatisfied customers. This duration is considerably above the industry benchmark of 30 days, reflecting inefficiencies that not only elevate operational costs by approximately15-20% but also diminish the firm's responsiveness to market shifts and opportunities for strategic sourcing. The results revealed that Recurrent Neural Networks (RNNs) are extensively used for time series analysis and forecasting, aiding in predicting future demand and supply trends at Petrozim Line Pvt Ltd. Feed-forward Neural Networks (FNNs) are employed for demand forecasting and price prediction, analyzing historical data to inform procurement decisions. Long Short-Term Memory Networks (LSTMs) are effective for long term trend forecasting and inventory management. The results showed that Recurrent Neural Networks (RNNs) have a statistically significant positive effect on lead time reduction (estimate = 0.987, p = 0.025). The positive estimate suggests that higher usage of RNNs is associated with greater lead time reduction. The effects of other ANN forms, such as Feed forward Neural Networks (FNNs), Long Short-Term Memory Networks (LSTMs), Autoencoders, Generative Adversarial Networks (GANs), Convolutional Neural Networks (CNNs), Radial Basis Function Networks (RBFNs), and Deep Belief Networks (DBNs), are not statistically significant (p > 0.05). The study results indicated that the most significant challenges identified at Petrozim Line Pvt Ltd are the high implementation costs, data quality and availability, integration with existing systems, algorithm complexity, and resistance to change. The study revealed that the recommendations for improving lead time reduction in procurement processes at Petrozim Line Pvt Ltd emphasize several key strategies. Implementing advanced analytics and forecasting tools, including ANNs, to predict demand and optimize inventory levels is highly effective. The study recommended that Petrozim Line Pvt Ltd should expand the implementation of RNNs across all procurement processes.

Date

JUNE 2024

Publisher

BUSE

Keywords

Artificial Neural networks

Procurement Processes

Case study

Petrozim Line Pvt Ltd

Zimbabwe

Supervisor

DR CHIGUSIWA