There is no doubt that AI/ML processes are highly effective at understanding, cleansing, forecasting, and predicting large and complex logistics datasets, making them invaluable for generating the INPUT data needed for system-based modelling. HOWEVER, AI/ML should not be seen as a replacement for these established and critical analytical processes. While the potential benefits of AI/ML are clear, excessive reliance on it can be just as detrimental as underutilisation.

AI/ML can be used to better understand demand patterns and the impact of external factors on those patterns. This can lead to accurate stock forecasting and optimisation; however, this capability is limited to steady state supply-related aspects and cannot fully analyse or optimize entire Product Support Networks over time. Given this, should we rely on complex statistical learning processes when simple, well-established mathematical algorithms are already available within proven Support Modelling and Analysis Tools and can deliver more knowledgeable outputs?

TFD Approach and Methodology

TFD Products and Services, focus on the optimisation and availability of Parts, Units and Assemblies that require Maintenance and Overhaul, and whose Repair Turnaround Times directly impact End System/Asset/Platform availability. Common metrics such as Fill Rates for Consumables, are easily and cheaply addressed by Kanban push supply and therefore need less analysis which can be delivered by AI/ML if desired. Consumable demands are relatively easy to predict using Establishment Variation Factors already embedded in reprovisioning algorithms. The TFD methodology, ensures the best path to successfully delivering ‘Happy Systems’, rather than ‘Happy Shelves’.

Our tools and techniques are most relevant in Product Breakdown Structure maintenance activities, particularly when events require external Repair Loops affected by various Administrative and Logistical Delays. If no Repair Loops are established, all stock is considered consumable and discarded at the point of removal. In contractual agreements, where spare parts are pooled or managed by a Service Provider, customers assume an unlimited supply for the agreed contract price. The service provider must then manage fluctuating demand from multiple customers and prioritize supply, based on the price each customer pays and the penalties for underperforming. This can lead to suppliers determining who receives stock from limited inventory to minimize their own costs and maximize profits, rather that ensuring customer satisfaction for all. With effective modelling and analysis of the associated support requirements at the earliest opportunity, evidence and knowledge can be established to support Supply and Resource Targets and the associated PIs in a Support Contract. This will result in correct volume of Spares and Resources being available, in the right place, at the right time to minimise maintenance delay, ultimately delivering the desired level of Asset/System/Platform Availability for the lowest cost possible.

Support is not just about spares

Operational Availability is determined by the full spectrum of IPS resources and their associated constraints. A repairable part is no use if it is stuck in an endless repair loop, or there are insufficient Facilities, Tools, or Personnel to conduct the repair or replacement. To complicate things further, there are often significant data misalignments between Asset Management IT systems and Inventory Management/Logistics Information Systems. These data gaps make it impossible to effectively correlate spare parts and labour costs with Maintenance Events. The data cannot be easily cleansed and analysed through traditional methods, and while AI/ML can assist, the volume of data (particularly for Repairable Parts) may be too limited to effectively train the algorithms, and implementing such solutions would require a considerable amount of time and data which may far exceeds what is available. The only truly effective path to Optimising Spares and other Resources, is through dedicated Support Modelling and Analysis Tools that use Marginal Analysis within the built-in algorithms – This analysis cannot be achieved within Spreadsheet Formulas, ERP Systems or other predictive procurement IT Solutions.

So, when is AI/ML acceptable

While there is no doubt that AI/ML techniques have an important role to play in the Supply Chain, offering capabilities that can significantly outperform human efforts in data conditioning and historical data analysis, we must also recognize their limitations. Predicting replacement of lifed items and scheduling preventative maintenance is relatively straightforward with AI/ML, however planning resources for Corrective Maintenance is much more complex and requires proven algorithms and a ‘Bottom Up’ modelling approach, which AI/ML alone cannot fully address. Once a baseline is established for your Optimised Support Network, AI can then be utilised to Simulate potential deviations from that baseline – this will maximise the Strategic Planning value of the Tools and Techniques, ensuring that your organisation can plan for any perceived eventualities in the Life Cycle of your Product/Asset/Platform.

Delivering Optimised Support Solutions to your Organisation

TFD can provide the combination of Support Modelling and Analysis Tools with suitable experienced personnel, to maximise the effectiveness of your Total Support System and not just Supplied Parts. We can deliver this on a Consultancy basis, or better still, we can train your own Product Support Team to use the Tools and Techniques, allowing you build your own ‘in-house’ enduring modelling and analysis capability.

Do MORE for LESS with SMARTER OWNERSHIP