Controllability and Observability in Power Distribution Systems

DERs present a golden opportunity to improve the resilience of the distribution system. This necessitates an intelligent management system previously employed only at the transmission level. The goal is a Distribution Management System (DMS) that will optimally utilize DERs to improve outage prevention and management. The DMS will rely on metrics for controllability and observability, dual concepts from control theory that serve as vital signs of sorts for the network. Our ongoing project in the AAC laboratory concerns the development of these metrics that adapt the definition of controllability and observability to the quasi-static distribution system. Supported by the US DOE Office of Electricity Delivery and Energy Reliability, our research investigations are focused on the development of a new framework for DER-based distribution system management. Highlights of our research are given below:

The notion of observability is a measure of how well internal states of a system can be reconstructed using a given set of measurements. In order to develop the observability analysis component of DMS, we have extended a transmission system observability test from the literature to the distribution system.The original algorithm was based on the canonical linear decoupled power flow model which is ill-suited to the distribution system where line resistance to reactance ratios approach unity.Instead, the test has been modified to accommodate the three phase linear coupled power flow model that was developed recently for the distribution system.To accompany the new distribution system observability test, we have developed an observability metric, based on the smallest eigenvalue of the gain matrix, that captures the visibility of the network based on the given measurements. Work is ongoing to exploit correlation between loads and DERs to improve observability. Techniques from parameter estimation theory may be leveraged for this problem.

The notion of controllability is a measure of how well internal states of a system can be controlled using a given set of inputs or actuators. In this case, the actuators are DERs as well as conventional control mechanisms such as tap changers and capacitor banks. A controllability metric is currently under development based on the voltage-current controllability index (VICI) from the literature.


Interdependent Natural Gas and Electricity Infrastructures

One of the fastest growing consumers of Natural Gas (NG) is the electricity sector, through the use of NG-fired power plants. With increased renewable energy generation, natural gas power plants have become key players due to their relatively low fuel costs and ability to ramp in response to changes in wind and solar resources. As a result, NG and electricity networks are increasingly linked and interdependent. Our lab has led research projects examining this interdependency, developing tools to improve the operation of these infrastructures. Two such tools are described in more detail below:

Reliability Contracts Between Renewable and Natural Gas Power Producers

Reliability Contracts Between Renewable and Natural Gas Power Producers

Renewable power adoption has required policies that protect intermittent generators, such as wind and solar, from system-level costs of resource shortfalls. It has been shown that if renewable generators were to accommodate these costs in energy market settlement, significant renewable generation curtailments would ensue, especially as the penetration of renewables grows. Based on the current evolution of policies towards unmet commitment penalties for intermittent generators, we developed a reliability contract between a renewable power producer (RPP) and a natural gas power plant (NGPP) where the NGPP fulfills the RPP unmet commitments in low resource scenarios. We analyze the contract against a baseline scenario where the RPP faces the energy market shortfall penalty, deriving optimal commitments and a condition where the adoption of the reliability contract increases social welfare. Using real data from a RPP-NGPP pair in Northeastern United States, the contract is shown to improve renewable utilization, increase the profits of both partners, and decrease total unmet commitments through the introduction of a lower-cost alternative to the shortfall penalty.

Modeling Renewable Generation Impacts on Natural Gas Markets

Computational tools, in the form of price regression models and market auction simulations, were developed to evaluate the impact of increased renewable adoption in New England’s coupled electric-natural gas system. Natural gas infrastructure parameters such as pipeline capacity were compared to forecasted requirement values for scenarios across varying levels of renewables, climate change impact and macroeconomic indicators. Both physical and economic system domains were studied, with a focus on developing recommendations to more efficiently manage the joint electricity-natural gas system and accommodate the growth of renewable generation. One of the main results from our simulation work is that increasing renewable adoption can increase natural gas market prices due to the increased variance in demand for natural gas demand, product of the increasing variance in renewable power output.


Dynamic Market Mechanisms for Wholesale Energy and Regulation Markets

Transactive energy control Renewable energy resources (RER) generate intermittent power which, along with fast variations of the electricity demand, cause power systems conditions to vary more frequently. This necessitates dynamic tools for analysis and control of power system operations and for market design, that can align the timescales of decision making and control processes with the natural timescales in which power systems’ conditions evolve. For this purpose, a series of dynamic market mechanisms (DMM) have been developed in the past decade in our group and are currently under development. These DMM realize dynamic solutions to different variations of optimal power flow problems in a market setting where, instead of solving the optimization problem at a single instant based on predicted conditions for a long horizon, they continuously utilize more frequently updated information about the systems’ conditions as it becomes available, providing new optimal solutions to these problems that correspond to the updated conditions. The DMM approach has been extended in many directions with its initial versions realizing solutions to optimal power flow problems for real-time wholesale markets to more recently realizing solutions to the optimal frequency regulation problem. DMM enables optimal and dynamic scheduling and utilization of resources in the fast power systems timescales, leading to improvements in economic efficiency and the efficient integration of renewable power.

Our recent work has addressed the following topics:
  • Stability of DMM
  • A Hierarchical Transactive Control Architecture for Renewables Integration
  • Integration of Demand Response in Electricity Markets
  • A Dynamic Regulation Market Mechanism leading to an optimal AGC and reduced make-whole payments

Co-Design of Wide Area Control and Monitoring of Power Grids

We address the problem of wide-area control of power systems in presence of different classes of network delays. We pose the control objective as an LQR minimization of the electro-mechanical states of the swing equations, and present an arbitration approach by which the flexibilities of the communication network such as scheduling policies, bandwidth, etc., can be exploited to design a delay-aware state feedback control law.

The delay-aware strategy is based on a sparse and distributed optimal control strategy. Sparsity is introduced in the underlying communication network on the basis of dominant participation of the state variables in the inter-area oscillation modes that decides the necessary generation units that need to communicate and included in the control design. In addition, the controller accommodates large network delays that are of values four to five times greater than the sampling period. A virtual sparsity concept is introduced to accommodate these delays by zeroing out the gains that correspond to measurements that are yet to arrive. Results are verified through a simulation study of the IEEE-39 bus power system model, where it is shown that with 86% less communication channels, we can obtain nearly 89% of the performance compared to the case with a non-sparse controller.

iTEACH Architecture for Control of Large Scale Distribution Grids

Given the large dimension of nodes in a distribution grid, a distributed energy management system that manages the huge volume of information and makes decisions that optimize global objectives and manages local outcomes will necessarily have to be hierarchical. This hierarchy will not only have to determine a spatial partitioning of various nodes but also a time-scale partitioning of various decision making. A new concept that’s showing a lot of promise is Transactive Control where intelligent agents such as flexible loads and dispatchable generators throughout the network negotiate the economic contracts to determine the price, quantity, location, and time of delivery of electric energy, thus balancing supply and demand in near-real time. The timescales involved in these negotiations (30 seconds to 5 minutes) compared to the timescales of primary controls of the distributed generators (milliseconds) endears Transactive Control to a hierarchical architecture. Transactive Control provides a significant potential for the active coordination of several DERs through suitably designed incentives for DERs to participate in economic transactions, and ensures reliable power delivery. Our lab is currently developing the iTEACH (integrated Transactive, Efficient, Actively Coordinated, Hierarchical) control architecture that will enable high penetration of penetration, increase the grid efficiency and provide a guarantee of system reliability to the end-user. To this end we are deploying a number of tools including Dynamic Market Mechanisms, hierarchical systems theory, optimization methods, nonlinear systems and control, lessons learned from industry experts, national labs, and pilot projects for high levels of renewable integration.

Wind Turbine Control

Adaptive Voltage Control of DFIG in Weak Grids

Wind turbines typically convert the wind energy into electric energy using induction generators. At every wind speed, the energy captured from the wind depends on the speed by which the blades rotate. Therefore, in order to capture the maximum available wind energy at different wind speeds, the rotation speed must be appropriately matched to the wind speed. This can be accomplished by regulating the electric power generated by the generator to be a function of the rotation speed. Earlier wind turbine technology did not provide this capability, which made the wind turbines optimal only at a certain wind speed. A more advanced technology, called doubly-fed induction generator (DFIG), does provide this capability. The wind turbine generators consist of a rotor that is mechanically connected to the blades through a gearbox, and a stator that is electrically connected to the electric grid. Power regulation in DFIG is achieved by controlling the voltage applied on the rotor circuits, whereas before these circuits were short-circuited such that no voltage was applied. For a given rotation speed, the power generated at the stator in steady state is a product of the voltage applied on the rotor and the voltage applied on the stator. Larger generators in the vicinity of the wind turbine can regulate the voltage applied on the stator such that it is always at its nominal value regardless of the power generated by the wind turbine. In this case the power generated by the wind turbine becomes a linear function of the voltage applied on the rotor. The widely used proportional-integrator (PI) controller is then able to regulate the generated power to the desired value. However, major wind resources can be found very far from the load centers, and the large generators that accompany them. The long transmission connecting the wind turbines to the main grid prevent the regulation of the voltage applied on the stator by external generators. The variability of this voltage, as a function of the power generated by the wind turbine, must then be taken into account.

We formulate a general control problem in which the output signal to be regulated is a nonlinear function of the state. We assume the parameters of this nonlinear function are unknown, but that both the state and the output signal can be measured. We then design a new controller that simultaneously estimates the parameters of the nonlinear function and drive the state such that the output signal is regulated as desired. In the manuscript listed below we prove that in the absence of noise, the output signal is indeed regulated as desired. We also describe in this manuscript how this controller can be applied to DFIG control in weak grid conditions, present its advantage over a PI-controller, and argue its robustness to measurement noise through simulation.

Publications:

Power Distribution Systems

Observability in Distribution Networks

The notion of observability is a measure of how well internal states of a system can be reconstructed using a given set of measurements. In this work, we derive necessary and sufficient conditions for observability in a power system. Deriving sufficient conditions for observability is quite difficult and algebraic observability is often used as a surrogate tool for observability. We show that algebraic observability is necessary but not sufficient for observability. It is also shown that standard measurement sets of at least one voltage measurement, and paired active and reactive power measurements may lead to unobservability for certain measurement configurations. Using a nonlinear transformation and properties of graph theory, a set of sufficient conditions are derived for observability. These conditions are shown to be dependent on the topological properties as well as the type of available measurements. The efficiency and robustness of the proposed approach is also discussed. The proposed method can be utilized off-line as a planning tool during the initial stages of measurement system design as well as on-line prior to state estimation.

Topology Identification in Distribution Networks

The electric grid is divided into high-voltage transmission networks and low-voltage distribution networks. Transmission networks carry the electric power from the major plants to the load centers. Distribution networks then distribute this power to the residential and commercial consumers. Because so many depend on the power flowing through every node of a transmission network, these networks are highly monitored and controlled remotely from a control center. Distribution networks, in contrast, are scarcely monitored and are managed manually by sending crews to reconfigure them. Until now, and despite having outages that sometimes take hours to identify and restore, there has been little demand to change the way distribution networks are operated. A major reason for this is that until now power flow in distribution networks always flowed one way from the substation, the point of connection to the transmission network, to the consumers. With the expected proliferation of distributed energy resources (DER), based mainly on wind and solar, we expect this to change. As more DERs are introduced, faults and subsequent line disconnections will affect not only those downstream but also those upstream. These networks, which traditionally were operated radially, may need to start to be operated with topologies that include loops in order to minimize power losses. All of these necessitate operating distribution networks more like transmission networks. However, the fundamental economics of distribution networks remain, and prevent equipping distribution networks with redundant real-time measuring capabilities covering every node in the network. Recognizing that the full state estimate required for transmission networks may not be required to perform the new tasks required in distribution networks, we ask whether we can add a few more sensors in order to provide just the information needed for these tasks.

Our recent research has focused on the problem of circuit breakers status detection in non-radial distribution networks. We ask whether we can detect the correct status in real time and with high probability, but without equipping every breaker with a sensor and a transmitter. We follow a state estimation approach, but since we do not have sufficient real-time data for state estimation, we complete the missing information using historical data. Because predicting the present conditions from historical data is probabilistic in nature, this leads to a machine learning problem. We then compare two tools in machine learning, namely the maximum likelihood (ML) and support vector machine (SVM), and find the former to perform better. We also provide a computationally efficient method to predict the success rate given the topology of the network, the location of the circuit breakers, and the placement of the few real-time sensors.

Publications:

Energy in Propulsion Systems

Energy efficiency is impacted dominantly by active control in several propulsion-related problems. Our research has focused on model-based active control methods that have led to energy efficiency via noise control. Key ingredients of our approach include reduced-order modeling, actuator design, adaptive control design, and numerical and experimental validation. Some examples are described below.

Combustion Systems

Our goal is develop a systems framework for analyzing and synthesizing an actively controlled combustor with optimal behavior in which active control is incorporated as an integral part of both the hardware and the software used to manage the different functions of the combustor. Several factors contribute to the functioning of a combustor, making its analysis and design a very complex task. These include, in the case of liquid fueled combustor, the liquid fuel injection, atomization, dispersion, evaporation, mixing and chemical reactions, flame stabilization through recirculation, and acoustics-vortex interactions. All of these processes are dynamically coupled, and together they affect the flow dynamics. They also interact with the acoustic dynamics of the system in ways which are determined by the flow conditions and geometry. The introduction of active control is yet another complexity which has to be designed so that it functions synergistically with the combustion process since actuation, whether it is acoustically implemented, or via fuel injection, imposes different conditions on the combustion process. Moreover, since active control inputs affect the acoustic field, the flame stabilization zone, and other flow dynamics mechanisms, a controlled combustor behaves dynamically different from an uncontrolled one.

We have employed a systems framework for analyzing such an integrated multi-component entity as well as for designing for optimal performance. Several reasons can be cited for this: (a) Since the ultimate goal is to optimize multiple global objectives from the combustor such as stability, low NOx, and high efficiency, several input-output relations must be known a proiri or found online. (b) The combustor can be viewed as a system made up of several subsystems (see Fig. 1), each of which can exhibit a strong temporal behavior that can be interrogated individually. (c) Tightly knit coupling is present among these subsystems (thermoacoustic resonance is one example of such a coupling; the creation of secondary peaks while using active control is another) which warrants the consideration of the combustor as a whole rather than a treatment of its parts. A systems approach enables one to study how different subsystems interact with each other and how they can be designed optimally for realizing a high performance.

The results of this systems framework have yielded (i) reduced-order models of combustion dynamics under weakly turbulent conditions and high Damkholer numbers, (ii) reduced-order models under highly turbulent conditions and low Damkholer numbers, (iii) model-based control strategies that have yielded optimal performance with an order of magnitude improvement in the pressure suppression in laboratory-scale and mid-scale rigs, and (iv) adaptive and nonlinear control strategies that result in enhanced performance in the presence of parametric uncertainties and time-delays.

A 75 kilowatt combustor has been designed and built to experimentally validate the combustion models and model-based control strategies developed in the AAC Lab. The experimental setup consists of a backward facing step-stabilized combustion tunnel, which can be operated as either premixed or nonpremixed, or with liquid fuel injection anywhere in the vicinity of the step, from a single hole, multiple holes or a slot. The combustion process is optically accessible through side windows and a variety of measurement and diagnostic techniques are available to interrogate the experiment. A Moog direct-drive valve is used as an actuator to pulse the secondary-fuel into the combustor. The facility is sufficiently flexible to allow for different test sections, different fuel injection locations and configuration.

A hallmark of unstable combustion is large pressure, heat release, velocity, and temperature oscillations. The backward step combustor is victim to the same phenomenon, as shown in the movie below where the light intensity of the flame is measured using a CCD camera operating at 1000 frames per second, indicating large temperature oscillations. The amplitude of the corresponding pressure oscillations is of the order of 5% of atmospheric pressure. Model-based feedback control strategies are being designed to reduce these oscillations.

The following movie shows active combustion control in a 85kW LPP(lean premixed prevapourized) combustor. Control was achieved by modulating the fuel flow rate using a Moog DDV valve in response to a measured pressure signal. The feedback control is an adaptive PosiCast controller which only requires the total time delay between actuation and response to achieve control. The algorithm achieves a reduction of up to 30 dB on the primary instability frequency. This performance was an improvement of 5-15 dB over an empirical control strategy (simple time-delay controller) specifically tuned to the same operating point. The experiments were performed at the University of Cambridge under the direction of Dr. Riley and Professor Dowling.

Supersonic Impinging Jets

Supersonic impinging jets produce a highly unsteady flowfield leading to a noisy environment with high dynamic pressure loads on nearby surfaces. In prior work, it was demonstrated that microjet injection along the circumference of the main jet nozzle directly into the shear layer of the main jet disrupts the feedback loop inherent in high-speed impinging jet flows, thereby significantly reducing the adverse effects produced. The microjet action was due to steady blowing whose flow rate was either uniform along the circumference or varied using the eigenmode of the flow. We are developing novel closed-loop control strategies for articulating the microjet pressure is suggested, in order to maintain a uniform, reliable, and optimal reduction of these tones over the entire range of operating conditions. Experimental results from a STOVL supersonic jet facility at Mach 1.5 show that these strategies lead to 8-10 db reduction, compared to an open loop one, at the desired operating conditions. More recently, a pulsing microjet injection was used as a new control scheme to retain a uniform suppression of impinging jet noise. Through this method injection, a fairly good amount of noise reduction was achieved using 42% of the mass flow rate that led to the same level of noise reduction using steady microjet injection. As the duty cycle was increased, it was observed that the pulsed injection completely destroyed the distinct impinging tone for almost all heights. A systematic control design using these novel pulsing microjets is currently being explored.

Blade Tonals in Underwater Vehicles

Modern SISUP, swirl inducing stator upstream of rotor, underwater vehicles consist of a torperdo-shaped body with a row of stators (fins) mounted upstream of a rear-mounted propeller. The purpose of these stators is to pre-swirl the flow to counteract propeller reaction torque as well as improve propeller efficiency. Unfortunately these stators also create a downstream velocity or wake deficit due to surface drag. This means that propeller inflow becomes non-uniform due to the spatial distribution of stators. As a propeller blade passes through regions of varying inflow velocity its effective angle attack changes and it experiences unsteady forces. These unsteady forces are a major source of directly radiated underwater noise. Each blade geometry results in a specific spectrum of noise, and therefore generates a specific "blade tonal" of the radiated noise

The goal of this research is to use active control to modulate a control surface in order to suitably modify the tonal. In this case a biologically inspired method of tail articulation, consisting of a hinged stator trailing edge, is used to modify propeller inflow. The flapping stator induces a point circulation which convects downstream towards the propeller. Experimental work has been carried out at both low and high Reynold's number to observe the effect of tail articulation on the stator wake.

Small-scale experiments carried in a low speed open-channel water tunnel at MIT and high-speed experiments carried out in a closed-channel water tunnel at the Newport Naval Undersea Warfare Center (NUWC) in Newport, RI, show that tail articulation can have a beneficial effect on the unsteady flowfield behind a stator. The image below shows the experimental setup used to obtain Laser PIV data for high-speed flow field observation.

The experiments showed that the most efficient tail articulation seemed occur at certain Strouhal numbers (St), which is a non-dimensionalized frequency of the flowfield. At low St, the stator generated vortex sheet moves in a quasi-steady fashion with tail articulation. At moderate St, the vortext sheet begins to roll up, and at high St, the vortex sheets roll up into tight discrete vortices of alternating direction.

A reduced order simulation was created to capture the major flow-field phenomena observed. A simplified 2-D non-lifting propeller blade was incorporated into the simulation to observe the effects of tail articulation on propeller forces. The simplified simulation allowed a parametric study to find the optimum noise reduction operating parameters. A 3-D propeller simulation (PUF) was then used to quantify any possible reduction in blade tonal noise. The simulation studies show that a 5dB reduction in radiated noise was possible. Current research is focused on the quantification of the effect of tail articulation on unsteady forces produced on a propeller downstream of the tail.