CO2 relevant optimization with machine learning and cloud technology
EU regulation and changing market demand have made carbon dioxide emissions a significant cost driver and influential factor in today’s decision-making processes, especially for automotive OEMs. Deloitte Carbon Analytics is a comprehensive simulation and optimization solution based on machine learning and cloud technology that helps them report and predict CO2 relevant KPIs as well as steer business considering CO2 related legal requirements.
In response to climate change, the EU has passed laws aimed at reducing industrial emissions of greenhouse gases, with the eventual goal of becoming carbon neutral in the future. Carbon dioxide (CO2) which is emitted by burning fossil fuels is one of the most important factors for human-induced climate change. That is why the EU has enacted legislations which force automotive OEMs to preferentially sell CO2 reduced vehicles. These regulations include CO2 targets for passenger cars and light commercial vehicles registered in European countries. Vehicle manufacturers that do not achieve these CO2 targets have to pay 95€ per gram per kilometer for each registered vehicle. According to the latest studies, the majority of the biggest automotive OEMs will miss the targets by the year 2021 and will have to face penalty payments of billions of Euros.
EU regulation and the shift of market demand towards CO2 neutral products have made carbon dioxide emissions a significant cost driver and influential factor in today’s decision-making processes, especially for automotive OEMs. On the other hand, manufacturers are lacking in insights towards their current and future CO2 emissions of their products and the correlation between CO2 and business steering key performance indicators.
With Carbon Analytics, Deloitte helps in mitigating costs resulting from penalty payments due to EU regulations for automotive OEM clients by providing deep insights into current and future CO2 emissions and their correlation to key performance indicators. These deep insights are getting integrated into both existing sales plan and strategic decision-making processes to add value from the first MVP onwards: Executives, managers and sales planners can simulate impacts on gap to CO2-targets and group-wide profitability in various scenarios ranging from “Brexit” to achieving eco-innovations, varying registration ratios and deviating fleet models. Furthermore, Carbon Analytics calculates optimal fleet model structures with regard to market demands and production capacities while taking into account the strategic goals of the manufacturers. The optimization of fleet variants towards market demands and production capacities finally allows decision makers to steer their region-crossing market penetration which adds significant business value.
Deloitte has already designed and implemented CO2 prognostics and reporting solutions for automotive OEMs in the past. Based on the industrial knowledge a holistic analytics concept was developed which enables simulations and optimizations in order to report and predict CO2 relevant KPIs as well as to steer business while considering CO2 related legal requirements. This analytics concept has been implemented end-to-end in a showcase based on carbon related functional and data models. These models can be reused as accelerators to speed up the first valuable carbon analytics deliverable and adapted to any industry. Beyond generating first valuable insights quickly, the recommended multi-cloud architecture allows customizing the final solution towards individual requirements in a flexible way. Extending analytics use cases won’t be an issue since storage and CPU resources can easily be extended. Experts in cloud-based frontend and backed technologies as well as the expertise in visualization, machine learning and algorithm optimization ensure solutions in accordance to the latest technologies of predictive and prescriptive analytics.
Deloitte Carbon Analytics is a comprehensive simulation and optimization solution that helps clients to report on their current and predict on future CO2 emission as well as to steer the group-wide fleet. Therefore, both a functional-model and a data-model has been developed which allows calculating the average CO2 emission of the entire fleet based on predicted sales plan figures. Thereby registrations are forecasted and regulations such as WLTP, super-credits, phase-in effects, eco innovations and dynamic CO2 targets are considered. The resulting penalty payments are taken into account in the overall profitability management as another cost factor. Production costs and market specific prices and taxes complete the perspective on the overall profitability.
Besides its predictive capabilities, Carbon Analytics offers prescriptive optimizations. By integrating the market demand and production capacities an optimization engine provides an optimal fleet setup concerning individual CO2 targets and strategic goals. Clients can set up specific CO2 targets in the future and strategic goals such as revenue, cost or profitability.
The results of both predictive and prescriptive simulation runs are finally consolidated and prepared for further reports. These reports consist of holistic side-by-side visualizations for both simulation runs, as well as drill-down functionalities from regions through countries down to vehicle models.
Technology, Methods and Algorithms
Carbon Analytics is a multi-cloud solution based on latest technologies designed for predictive and prescriptive problem solving. The backend stores and integrates relevant data sources in a dedicated cloud platform. Additional simulation engines empower modeling the what-if scenarios while executing prediction and optimization runs. In order to predict sales plan figures, machine learning algorithms are leveraged that apply the optimal time series analysis given the historical data. Data Mining algorithms additional help the client to foresee future vehicle registrations. Cluster algorithms reveal correlations between CO2 emissions and KPIs in order to derive consequent marketing activities in a more efficient way. In order to calculate the optimal fleet setup, linear programming algorithms and libraries are used. Visualization of side-by-side scenarios and various drill-down functionalities as well as the calculation of key figures are realized with a live connection from the cloud-based backend to another cloud dedicated front-end for visualization purposes including cloud-based reporting.