A Stochastic Resource-Sharing Network for Electric Vehicle Charging
summary
Models electric vehicle charging as a queuing network on a distribution tree, using a linearized DistFlow approximation plus utility-maximizing allocations (mainly weighted log utility) and a fluid scaling to get a deterministic invariant point. Claims: the steady-state uncharged fraction reduces to a single convex optimization (knapsack-style in examples), processor-sharing/Erlang limits appear in special cases, and fluid predictions match AC and stochastic simulations within ~10% for tested two-node cases (K≈50). Motivation includes a 2025 scenario where a single Tesla causes a ~5 V drop and a network fault.
AI-written summary · about
chapters (13)
- 0:00Intro
- 1:00Background
- 2:10Charging Options
- 4:15Charging Stations
- 5:40Experiment
- 7:50Model
- 13:25tractable load flow model
- 16:35static load flow model
- 22:05other questions
- 26:15fluid scaling
- 33:00Setting Weights
- 33:49Resource Allocation
- 36:38Urban Math
related talks
53mOptimal Power Flow: Relaxation, Online Algorithm, Fast Dynamics
21mChatty Tenants and the Cloud Network Sharing Problem
49mDistributed Control for Demand Dispatch
12mTwenty Years After: Hierarchical Core-Stateless Fair Queueing
25mApproximating Fair Queueing on Reconfigurable Switches
01h 07mIndustry Presentations
29mFirst Order Methods for Distributed Network Optimization
16mCredence: Augmenting Datacenter Switch Buffer Sharing with ML Predictions