As storage, particularly battery storage, becomes cheaper, utilities are transitioning to a new model of operation, incorporating more renewables, storage, and EVs will play a part in this. The downside is greater complexity in a bi-directional grid.
Everything to do with EVs is dropping in cost, including batteries and the price of the vehicles themselves. More fleet operators are seeing the advantages of EV delivery vans, taxis, buses and trucks, compared to diesel-powered ones: fuel costs vastly reduced, less noise and lower pollution. New challenges have to be faced: these include managing widely-varying costs of energy from the grid, sometimes even negative pricing to soak up excess energy production, so the user is actually paid to take energy. Limitations on the local grid connection; the preference for locally-generated energy – for example microgrids – over national sources. As well as upskilling depot staff to enable them to make use of variable prices, while still keeping the vehicles to their tasks. A further complication is that most fleets are mixed: ICE vehicles are working in tandem with EVs, as older transportation is upgraded.
Limitations of the grid are a significant factor. If the depot grid connection is insufficient, then compromises will have to be made. This could limit the charging rate at certain points during the day. This may be alleviated with local battery storage that is ‘filled up’ with a combination of grid and locally-generated renewable energy from solar photovoltaic (PV) panels. Predicting local PV generation, EV energy needs, and local battery state of charge is a complex and continuous prediction, optimization, and control operation best suited to automated techniques able to solve for multiple objectives with complex data inputs. Furthermore, the advent of V2G services introduces the added complexity of bi-directional power flows at the connection to the utility grid.
If large scale electrification of vehicle fleets continues, then utilities and grid operators will be faced with sudden spikes of demand, as large numbers of vehicles are charged simultaneously when they return to the depot. In order to mitigate this, utilities are already offering demand-shifting opportunities, where price is lower, rather than start costly grid upgrades. This will improve energy efficiency.
A combination of battery storage, PV panels, and intelligent energy management can ensure that costs, loads, and need to recharge vehicles for operations are in balance. There should also be costs and emissions savings by using electricity at the optimum price point.
The complexity of EV fleet operation optimization together with the deluge of available data make AI and machine learning solutions the best candidates for the job. They are swift to operate, in near real-time. They can cope with the rapidly-changing circumstances throughout the day.
AI-enabled energy management systems are best operated with a distributed system. This software architecture can use better computing resources to capture data and make decisions. The high-performance computing needed to build AI models and manage the network holistically can be performed centrally, in the cloud. Such an architecture is well-suited to tackling the challenges for the following reasons:
• Connects to the numerous different assets and provides system-wide coordinated optimized recommendations for optimal operation
• Operates in the cloud for maximum computing power, flexibility, and security
• Solves for multiple objectives in parallel, allocating different priorities to the different performance metrics
• Works with different hardware and management systems, to inter-operate with all standard fleet management infrastructure
• Learns as it operates, continuously improving performance and adapting to changes in the system without the need for human intervention or software adaptation
Â
Overall, EV fleets will need to interface with AI and ML systems if they are to ensure that the grid does not suffer overloads and outages from the need for country-wide electrification.
Â