Cyber-Physical Energy Systems: Focus on Smart Buildings

Presentation by Francis Usher

CS 525 Fall 2012

Monday September 19, 2012

Goals

● Examine challenges for energy systems● Virtual case study

– Implementation challenges in RL scenario– Estimated energy costs and prodction

● Thought experiment– How close can we get to self-sufficiency?

Motivation

● Buildings: 70% electricity, 40% GHG● Push for Zero Net Energy Buildings (ZNEB)● Coupled generation, sensing, & control

– Just one or two of these inadequate

Energy Metrics & Concepts

● Detailed breakdown of energy usage– “Mixed-use” building– Lighting, mechanical, machine-room, plug– Baseline load

● Large % total load● Unreasonably high

Experimental Design

● UCSD CSE building (mixed-use)● Normal weekday occupancy● Lots of permanent heavy-duty IT equipment● Large roof space● Climate control (overlooked)

Mitigation Approach

● IT Infrastructure (Somniloquy)● Lighting (Motion sensors & LED)● Generation (Solar PV)

– Other generation forms not widely suitable

Somniloquy

● People don't put their computers to sleep!● Must allow lightweight, always-on net apps● Pilot test on 30 CSE desktop users

– 60-80% savings

● Server estimate 50%– Workload consolidation– Virtual machine migration

Lighting

● Off-hours mitigation– Don't jeopardize productivity– 20% of baseline = safety critical lights– Other lights on motion sensor

● LED lights– Reduction to 1/3

Solar PhotoVoltaic (PV)

● Function of available roof space (2700m²)● AC conversion efficiency vs. requirements

– 15% (standard)– 25-48% (necessary with tracking)– 31-66% (thermodynamic limit)– 32% (laboratory multi-junction)

● Update– 30% (ZTJ space-application multi-junction)– 43% (laboratory multi-junction)

Main Results

● Breakdown of metered energy use– Baseline plug load (PC energy use) ~ 50%

● Savings capacity– Mostly IT, some light

● Comparison with PV generation capability– Contemporaneous PV technology inadequate

Future Work

● Real-world tests– Implement changes– Collect data– Fine-tune, tweak

● Analysis of cooling & ventilation optimization– Make climate control ~ actual needs– More efficient fans & pumps– Closed-loop zonal climate control– (assumed savings of 20%)