Dynamics AX proves to be a great platform for managing fleet predictive maintenance. Machine breakdowns cost construction companies time, money, and opportunity. These costs can be managed and minimized if they are planned days or weeks in advance---but how do we predict which machines will be in need of service?
Using Cortana Intelligence Suite (CIS) components, we collaborated with XAPT, a global Dynamics ISV, to build an application that predicts breakdowns based on machine telemetry and maintenance history data stored in Dynamics AX.
As a Dynamics AX ISV, XAPT sells a custom extension of Dynamics AX for the heavy machinery industry to support dealers in offering service contracts to their customers. Given a customer that has a large construction project that runs into many months, their practice is to sign a lease with the dealer that includes periodic service, in the hope of avoiding unscheduled break-downs to their heavy equipment. The current practice has been to conservatively estimate a “one size fits all” frequency of service as recommended by the manufacturer for this equipment.
With CIS we can predict equipment failures ahead of time, so the dealer can provide more competitive services by adjusting the smart date, the interval before service is required, matched to the prediction for this equipment. Instead of selling the customers canned service contracts based upon fixed service intervals, they can offer the customers customized service contracts with durations matched to the historical and real-time data about the equipment. This is important, because it helps the sales manager decide the optimally priced service maintenance contract -- and the customer saves money too.
Here, we see a chart of smart dates for a list of machines in service. The colored segments show remaining predicted time-in-service, as determined by their smart dates.
Cost Savings of Intelligent vs. Conventional Contracts
The cost savings depend on the time-value of operating the machinery balanced with the risk and cost of unplanned failure. Our model determines risk as a function of the total time the machine will be in operation. AX cost accounting tracks the value of operation and the cost of repair. Previous service records tracked by XAPT, combined with real-time telemetry from machines in operation in the field determine risk of failure over time in the future.
Optimizing predictive maintenance by predicting risk
Extending service intervals runs the risk of increased failure probability. Shortening intervals means possibly retiring functioning machines for service before its necessary. Risk predictors learned from data of equipment failure over time are used to optimize this scheduling trade-off by setting “smart dates” for maintenance.
Smart dates are adjusted as new telemetry information changes risk estimates. Dealers manage their contracts in order to target service maintenance more effectively by both adjusting the initial service contract length to match known failure profiles based on historical data, and to accelerate service schedules during the contract based on telemetry when it suggests increased failure risk.
During use, if the equipment is predicted to be likely to break down then maintenance could be accelerated. And sadly, if there is an equipment failure out of contract, then the customer even could be presented with the presumptive contract maintenance cost in comparison to the costs unexpectedly incurred.
The smart date application runs in part in the Azure Cloud, and in part as an extension to the existing Dynamics AX on-premise application. This extension is accomplished by a custom AX job that uploads data to Azure cloud storage on a daily cadence. An Azure Data Factory then consolidates and routes the data to an Azure Machine Learning (AML) service where a series of statistical models are trained. Custom forms in AX then query risk estimates from the models as needed via standard web calls.
The solution we've shown here is a great example of how to extend the capabilities of Dynamics AX with the Cortana Intelligence Suite by building on an existing AX implementation. For additional examples of Cortana Intelligence based solutions, see this video from ThyssenKrupp and the Predictive Maintenance for Aerospace entry in the Cortana Intelligence Gallery. If your organization is considering a predictive maintenance solution, feel free to contact us for help.