Effective Spatial Decision Support for Charging Infrastructure Planning

QC241112

Abstract

The transition to electrified road transportation is crucial for achieving sustainability goals and reducing greenhouse gas emissions. However, therapid adoption of battery electric vehicles (BEVs) depends heavily on the availability of a robust charging infrastructure. Effective charging infrastruc-ture planning faces numerous challenges stemming from deep uncertainties inherent in transport electrification. These uncertainties encompass aspectssuch as rapid technological advancements, the variability of technology adoption and behavioral changes, the shifting landscapes of regulations, policiesand subsidies, the variability in availability, cost, development lead-time for grid transmission capacity, real-estate, and related services, and the evolving market dynamics arising from competition. This thesis examines the complexities of charging infrastructure planning, addressing two critical knowledge gaps identified in the literature: the inadequate utilization of transport route information in charging network placement optimization and the lack of planning methods and tools that can help manage the uncertainties.

To this extent the research presented in this thesis, after a quick background on transport electrification needs, challenges, current status and future ambitions, analyzes charging infrastructure planning support needs with respect to the knowledge gaps. Specifically, through a clear example, the thesis argues for the need to adopt a demand-centric charging network design where the adequate use of transport route information to achieve logical network design objectives is undeniable. Moreover, quite logically, the thesis argues the use of a dynamic adaptive planning approach that requires interactive decision support tools to manage the deep uncertainties of transport electrification during the system transition. Then, through a broad and systematic literature review, the thesis establishes that the prior work has not fully addressed these planning needs.

To extend scientific knowledge on the adequate use of transport route information in demand-centric charging network design, the thesis formalizes a data-driven simulation based transport electrification scenario model and in it an incremental charging network placement optimization problem where both the model and the problem require the use of detailed information in the transport routes. Then, the thesis proposes a series of greedy network expansion based charging network placement optimization methods to tackle the combinatorial network design problem and the incremental planning support need. First a set of baseline methods are introduced that guide the exploration of the search space using inaccurate but easily pre-computable static demand proxies and attempts to correct inaccurate guidance of the proxies by enforcing spatial constraints on the network placements to avoid the demand losses within the network and increase the electric coverage provision of the network. Next, the thesis proposes the Route Based Network Demand (RBND) method that does not use approximations or heuristics but rather selectively recomputes via computationally demanding simulations the exact values of the objective functions for promising network expansion candidates during the search space explorations. Empirical evaluations assess the methods’ optimization quality, empirical optimality, sensitivity to model parameters, and runtime scalability. The results show that the RBND method outperforms the baseline methods in optimization objectives and, among a practically infinite number of possible solutions, identifies statistically provable near-optimal solutions within minutes.

To address the lack of planning methods and tools that can help manage the uncertainties, the thesis contributes to scientific knowledge in two ways.  First, the thesis proposes the parameter sensitivity analysis of optimized charging network placements and as a case study explores and aggregates the optimized network placement information for 324 combinations of 5 key transport electrification scenario parameters as an attempt to derive the likelihood that a given location is part of an optimized network and what is the average charging demand at that location across all tested scenarios. Notably, the sensitivity analysis methodology entails the selective evaluation and simulation of the entire Swedish road freight system with 10.5 million annual transport routes for the most promising few million charging network placements for each of the 324 transport electrification scenario parameter combinations. The results of this vast charging network placement- and scenario search exploration are presented in a single table and two maps that represent frequency- and the average demand distributions of locations in optimized networks. Second, motivated by the argument that charging networks are developed by individual economic actors with their unique opportunities, challenges, and strategies in a competitive environment, based on qualitative feedback from 33 stakeholder organizations, the thesis describes the design, the components, and the visual analytics features of a flexible, spatial decision support system that can support the dynamic adaptive planning in stakeholders’ competing and collaborative settings.

Together, these contributions provide a scalable and flexible framework to support the planning and deployment of sustainable and resilient charginginfrastructures, while addressing uncertainties and enabling dynamic adaptive planning to meet the evolving requirements of electric transport.

Finally, the thesis points to three distinct future research directions. First, the extension of the methodologies and systems is called for to provide plan-ning support for charging network designs that include a mix of electrification techniques, e.g., dynamic charging on electric roads and static chargingon stations. Second, with a similar methodology as the developed sensitivity analysis, the thesis highlights the importance of network resilience and callsfor the methods that evaluate and integrate network resilience in the designs. Finally, realizing that regardless of the sophistication of long-term strategic planning not all uncertainties can be mitigated, the thesis calls for methods and tools that can increase the operational and cost efficiency of charging networks and electrified transports on them.

urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-356237