Cristian Perfumo

Model-based feedback control of distributed air-conditioning loads for fast demand-side ancillary services

Load control (LC) of distributed populations of air conditioners (ACs) can provide effective demand-side ancillary services while reducing emissions and network operating costs. Pilot trials with ACs typically deploy model-free, open-loop strategies, which cannot deliver the full potential of LC as a network resource. Seeking more advanced strategies, much research in recent years has targeted the development of accurate models and LC approaches for this type of loads. Most existing approaches, however, are restricted to scenarios involving large numbers of ACs, which may not work in small populations, or require two-way communications with the controlled devices, which may come at high costs in widely distributed populations. This paper exploits a previously developed dynamic model for the aggregate demand of populations of ACs to design a simple controller readily implementable in such LC scenarios. The proposed feedback scheme broadcasts thermostat set-point offset changes to the ACs, and requires no direct communications from the devices to the central controller, using instead readings of total aggregate demand from a common power distribution connection point, which may include demand of uncontrolled loads. The scheme is validated on a numerical case study constructed by simulating a distributed population of ACs using real power and temperature data from a 70-house residential precinct, and is shown to deliver robust fast load following performance. The simulation results highlight the practical potential of the proposed model and feedback control scheme for analysing and shaping demand response of ACs using standard control techniques.

Model-Based Estimation of Energy Savings in Load Control Events for Thermostatically Controlled Loads

Load control (LC) of populations of air conditioners (ACs) is considered suitable to shift energy from on- to off-peak times, and track the intermittent power output of renewable generation. From a technical and economical point of view, it is paramount to quantify the amount of energy that can be saved by implementing these LC events. This paper proposes a new causal methodology to estimate such energy savings using a Kalman filter that includes a parametric second-order model of the aggregate demand of a population of ACs. The proposed methodology relies only on readings of aggregate electrical power at the feeder level and does not require historical load data, or a control group, and hence, it can be used where other methods reported in the literature are inapplicable. The proposed estimator is evaluated on a numerical case study that embeds simulated ACs in real power and temperature data from a 70-house residential precinct.

Controlling datacenter power consumption while maintaining temperature and QoS levels

The large amount of energy used by datacenters impacts both energy cost and the electricity grid. These issues can be mitigated by dynamically adjusting the power demand of datacenters. However, conflicting objectives have to be considered: workload and cooling can be dynamically reduced, but with a potential impact on quality of service or excursions beyond acceptable temperature bands. In this paper we use a multi-objective evolutionary algorithm to explore the trade-offs between power consumption, temperature, and quality of service when both servers and cooling are controlled holistically in a datacenter.

Reducing energy use and operational cost of air conditioning systems with multi-objective evolutionary algorithms

Air conditioning is responsible for around 60% of energy use in commercial buildings and is rapidly increasing in the residential sector. Although each system is individually small, the proliferation of air conditioning and the correlation of energy use with temperature is driving peak demand and the need for electricity distribution network upgrades. Energy retailers are now looking for ways to reduce this aggregate peak demand, leading to a tradeoff between peak demand, energy cost and the thermal comfort of building occupants. This paper presents a multi-objective evolutionary algorithm (MOEA) to quantify trade-offs amongst these three competing goals. We study a scenario with 8 air conditioners (ACs) and compare our findings against the case of having all ACs working independently, irrespective of global goals. The results show that, with statistically significant certainty, any run of the MOEA outperforms any scenario where the ACs function independently to keep a given level of comfort on a typical hot day.

Mapping the Effect of Ambient Temperature on the Power Demand of Populations of Air Conditioners

The direct load control (DLC) of populations of air conditioners (ACs) is a cost-effective demand-side strategy to manage peak demand and provide ancillary services to the electricity grid. Much research in recent years has focused on the formulation of mathematical models for the aggregate demand of such populations as a basis for DLC analysis and design. These models, however, are sensitive to ambient temperature, which is often assumed constant, and rely on the knowledge of thermal parameters that are typically difficult to obtain. This paper develops a new mathematical model to map the effect of ambient temperature to the aggregate demand response of a heterogeneous population of ACs. The significance of this new model lies in that it can be used to estimate the thermal parameters that characterize aggregate demand using available local weather and demand data, with no need for additional metering or direct engagement of grid users. While the estimation of these parameters is of independent interest in mapping distributed building energy performance over the grid, a key additional benefit is that these parameters fit existing models for DLC design, which provides the technical basis for practical model-based DLC of populations of ACs.

Holistic multiobjective planning of datacenters powered by renewable energy

Energy efficiency is a major concern to datacenter operators, because the large amounts of energy used by parallel computing infrastructures increases costs and affects the electricity grid. Datacenter power consumption can be reduced by applying intelligent control techniques to dynamically adjust power demand, but this is hampered by conflicting objectives. For instance, the workload can be controlled to adjust power, but at the expense of service quality. Or, the cooling infrastructure demand can be manipulated without affecting workloads, but at the risk of shifting the datacenter temperature outside the acceptable limits. This paper proposes a multiobjective, evolutionary approach to solving the problem of energy-aware task scheduling in datacenters. Our approach takes into account three problem objectives (power consumption, temperature, and quality of service) when both computing and cooling infrastructures are holistically controlled. We report the two best solutions to each of these problem objectives, as well as the selected trade-off solutions between them.

An analytical characterisation of cold-load pickup oscillations in thermostatically controlled loads

Large groups of thermostatically controlled loads can be controlled to achieve the necessary balance between generation and demand in power networks. When a significant portion of a population of thermostatically controlled loads is forced to change their on-off state simultaneously, the aggregate power demand of such population presents large, underdamped oscillations, a well-known phenomenon referred to by power utilities as “cold-load pickup”. Characterising these oscillations and, in general, the aggregate dynamics of the population facilitates mathematical analysis and control design. In this paper we present a stochastic model for the power response and derive simple expressions for the period and envelope of the oscillations.

Bayesian Parameter Estimation for Direct Load Control of Populations of Air Conditioners

Abstract Recent approaches for direct load control (DLC) of populations of air conditioners (ACs) to provide demand-side services in the electricity grid rely on mathematical models of the aggregate demand dynamics of these populations. These models can be parametrised by the physical characteristics of the ACs in the population, for example their thermal power. The knowledge of how their physical parameters are distributed in the population of real devices is instrumental in the analysis and implementation of controllers based on such models. For large populations, it is typically assumed that these parameters are stochastically distributed according to some probability distribution, e.g., log-normal, which has been effective in simulations. However, the identification of such distribution for a specific population remains an open problem for real-world deployments of DLC. This paper formulates a Bayesian framework for the state and parameter estimation of a previously developed input/output model for the aggregate demand response of heterogeneous populations of ACs. This framework enables us to assign a prior distribution to the parameters of the model, which is then updated using measurements of power demand data for the population to reach a posterior distribution that is more informative about the true value of these parameters. The framework uses sequential Monte Carlo methods, which are well-suited to existing high-performance computer hardware, and aims to provide a way to fill a gap between simulation and real implementation by validating posterior parameter distributions using real measurements. Simulation results indicate that our approach can successfully capture the values defining the distributions of physical parameters in a population simulated by 10,000 ACs with a standard hybrid dynamic model for each device.

A sensitivity analysis of the dynamics of a population of thermostatically-controlled loads

Load control of populations of thermostatically-controlled loads (TCLs) is considered a promising approach to match generation and consumption in electricity grids from the demand side. However, when these loads become synchronised they present a decaying oscillatory aggregate demand, which results in undesired power peaks. In this paper we describe the nature of these oscillations and develop a list of factors that determine its shape. We perform a sensitivity analysis which allows us to identify important relations between the physical parameters of the TCLs and their aggregate dynamics. Beyond describing fundamental behaviour, these relations can help develop and validate analytical expressions that facilitate control design, enabling the use of TCLs for demand response.

Modelling the aggregate demand response of a population of air conditioners to changes in ambient temperature

A substantial amount of research in recent years has investigated the direct load control (DLC) of populations of air conditioners (ACs) to provide demand-side services in the electricity grid. In many existing approaches, the control of aggregate power demand of these populations requires the knowledge of distributed physical parameters, such as the rated thermal power of the ACs and the thermal capacitances of the air-conditioned spaces. These parameters can be identified from DLC trials on a real-world population. However, such trials typically need to engage participants and fit their ACs with DLC-enabling devices for monitoring and control, which can be costly and may raise privacy concerns. This paper develops an alternative approach that allows the non-intrusive identification of distributed parameters for DLC. The proposed approach is based on a new mathematical model that describes the dynamic aggregate demand response of a population of ACs to changes in ambient temperature, rather than to a control signal. The parameters of the proposed model can then be identified from ambient temperature and aggregate demand data from sufficiently warm days, which are inexpensive to collect and do not need the direct engagement of participants in the target population. A key benefit of the new model is that its identified parameters also fit a previously developed dynamic model for DLC of aggregate demand of ACs, which completes a practical solution to model-based feedback control design for DLC in such populations.