My main interest is in studying the interaction between buildings and the energy system in the context of planning, design, and operation. The research projects I conduct and manage develop the underlying methodological approaches to empower policy makers, urban planners, building operators, and facilities managers to achieve climate change goals through data-driven techniques. These approaches are developed using knowledge of building physics, energy system dynamics, applied control theory, optimization, times series modelling, and machine learning techniques.
FlexTECC: Flexible Timing of Energy Consumption in Communities
This is my EPSRC Innovation Fellowship (EP/S001670/1, £496k) running from June 2018- May 2021. The aim is to develop hierarchical model predictive control strategies to enable collections of buildings with electrical building energy systems (those used for heating, cooling, and ventilation) to provide sustained demand reductions (reduction of x for t amount of time). The project has three components: theoretical development, performance evaluate in virtual test beds, and performance evaluation in a physical test bed of two test homes and a commercially operated building).
The simulation research, as reported in this paper, shows sustained demand reductions can be achieved under certain conditions, specifically during the mild swing seasons. However this reduction can come with a very large pre-peak in heat demand when using predictive control approaches. This peak, however, can be substantially reduced if the timing of demand reduction and selection of buildings to provide the demand reduction is centrally coordinated. In this work, full knowledge if each system was assumed. Ongoing work is looking to minimizing the amount of information exchanged to achieve the same results.
Ongoing works are developing experimental tests to achieve these results in practice using the Loughborough University test homes and a commercially operated building on campus.
Optimal Urban Scale Retrofit Analysis
Eighty percent of buildings in Western cities will still be around in 2050. Therefore to achieve the energy reductions required in the next thirty years will need to retrofit the current building stock. We, as researchers, can develop optimization approaches to determine the best set of retrofit options to minimize costs and energy consumption. The question is “how can we incorporate broader societal issues, such as equity, into the solutions that these optimization models can provide? “
Data-driven Community Building Energy Modelling
How much data do we need to identify detailed building physics and energy systems models that can accurately estimate hourly energy demand ? How detailed of a model do we need given the data we have available? This research is attempting to answer this question through a systematic high dimensional model selection approaches using detailed monitoring data from 50 buildings on campus.
Estimating & Mapping New York City’s Energy Demand
In this work, I combined public and opportunistic data sets through a statistical analysis to develop New York City specific estimates of annual energy end use intensities for different building types. The annual EUI’s were then applied to the floor area of NYC allowing us to map the building energy consumption of New York City. The interactive map of the estimates can be found here https://qsel.columbia.edu/nycenergy/ .
Estimating Occupancy Count in Buildings Through ICT Data Sets
Collecting data on building occupancy is crucial for understanding building energy consumption and enabling automated operations to minimize energy efficiency. However many buildings don’t currently have this sensing capability as the cost of the additional infrastructure is not seen as worthwhile. But is it possible to use information currently being collected by a building for general purposes to understand occupancy count, i.e. how many people are in a space?
In this project, we tested this hypothesis using an office building on campus. We collected data about the number entries and exits from the security access system, the number of mobile devices connected to wireless access points, and the number of active desktop computers in hourly intervals. We compared this information to a commercial video based occupancy counter to see if these data sources could indeed reflect occupancy count. We found that all data sources were able to pick up the occupancy peaks and troughs however the magnitude of the estimation was off by different factors. Therefore calibration of each data set would be needed. By testing in a variety of scenarios, we found that each data set could be modified to reflect occupancy count by simply multiplying by a constant factor and that it would take only 24 hours of data to estimate this parameter, although a week reduces the risk of picking a less representative day.
Estimating Marginal Emissions factors for New York City & New York State
The benefits of energy efficiency measures are often estimated with marginal emissions factors, with the assumption that changes to demand will most likely not affect baseload operation. The mix of generators used to provide electricity or that would be affected by changes in demand is determined by the local context of supply and demand. Therefore local estimates of marginal emissions factors as they are now and how they may change in the near future are important for making decisions on how energy efficiency measures will affect greenhouse gas emissions. In this work, using empirical data on the operation of hundreds of generators in New York state combined with a regional unit commitment problem, estimates of marginal emissions factors for New York City and New York State were made. The analysis indicates that marginal GHG emissions factors for NYC could reduce 30% in the next 10 years given the current plans for increased wind generation. This will have a major affect on the viability of combined heat and power projects from a GHG emissions point of view.
Best Operational Strategies for Combined Heat and Power Systems
If you want to add a building scale combined heat and power (CHP) system, how should you operate it i.e. determine how much electricity and heat to produce at any given moment, to reduce greenhouse gas emissions? In this simulation study, building scale CHP systems were sized and operated in typical multi-family residential, office, and hospital buildings in a variety of US climate zones, with a variety of emissions factors. The result showed that the optimal operation for most scenarios is to operate with an electrical load following strategy due to the high GHG emissions factors across most of the country. In regions with lower greenhouse gas emissions, a mixed strategy, trading off whether to produce heat or electricity first to meet demand is best.