Enhancing aftermarket demand planning with product-in-use data

Licentiate thesis, 2019

Enhancing aftermarket demand planning with product-in-use data

ABSTRACT

Aftermarket demand planning, consisting of forecast and known demand, is a critical activity for both the uptime of customers’ products and the supply chain-related costs. Traditional aftermarket forecasting methods use historical demand as the only input to the statistically based forecasts, usually combined with judgmental modifications. Since the underlying mechanisms of  customer demand are the need for maintenance or repair, these methods fail to deliver good forecast accuracy in various contexts, e.g. phase-in/out of spare parts or for spare parts with intermittent demand patterns

However, rapid increase in access to new types of data created by on-board sensors, improved algorithms for predicting future events with massive amount of data and increased Information and Communications Technology (ICT) capabilities are generating new and potential improved demand planning methods for aftermarket services.

The purpose of this thesis is to investigate how product-in-use data can be used in, and improve the performance of, demand planning processes for aftermarket services, by describing and explaining related performance effects and challenges.

The conceptual framework developed consists of three components (potential demand planning methods, challenges for implementation and development of these methods and confirmation of the results). Each component is attached to a research question which is dealt with in three separate studies. The two first studies are single-case studies which address the potential improved demand planning methods and challenges in aftermarket supply chain planning processes. The third study combined qualitative and quantitative methods, of which the latter was a regression method with exogenous variables (ARX).

Results show how product-in-use data is best utilised in planning spare parts with different attributes, e.g. different life-cycle phases and demand frequencies. Eight potentially relevant interventions, i.e. proposed methods, using product-in-use data in the demand planning process are identified. Challenges are explored in relation to the process complexity and data complexity of aftermarket supply chains, and underlying reasons and interdependencies are identified. Forecast accuracy is proven to be better for phase-in spare parts with medium/high demand frequency in a quantitative study.

The practical contributions of the thesis are insights how to enhance the aftermarket demand planning by new causal-based methods using product-in-use data, as well as awareness of the complexity and challenges of development and implementation of such data-driven processes. The thesis contributes to theory by proposing how product-in-use driven demand planning methods can be used and for which types of context the methods can create value. The process focus contributes to how to apply and develop methods for aftermarket causal-based demand planning