Tuning and Testing an Online Feedback Optimization Controller to Provide Curative Distribution Grid Flexibility
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Robotics Radio. Generated commentary on the latest robotics and control papers.
Rosa: We're looking at the paper: "Tuning and Testing an Online Feedback Optimization Controller to Provide Curative Distribution Grid Flexibility".
Dev: Here's the gist of the paper. Online Feedback Optimization (OFO) is presented as a new control method that steers power systems to an optimal solution of an optimization problem using minimal model information and computation power.
Rosa: This paper experimentally verifies that an OFO controller is a powerful tool to disaggregate flexibility requests onto FPUs while satisfying operational constraints inside the flexibility providing distribution grid.
Dev: Problem Definition and Methodology The core task involves defining an optimization problem whose solution provides setpoints to the Flexibility Providing Units (FPUs).
Rosa: The authors compare two fundamentally different ways to define this optimization problem and give a clear recommendation on which formulation is superior.
Title and authors: Rosa: Now that we know what it is about, let’s look at the specifics of what this paper, "Tuning and Testing an Online Feedback Optimization Controller to Provide Curative Distribution Grid Flexibility," actually does. They compare two ways to define the optimization problem: a cost approach and a constraint approach.
Dev: That comparison is important because they actually recommend the constraint approach because it leads to lower losses in the grid, which is a big deal for efficiency, and it's more versatile because you can use its cost function to parameterize certain behaviors.
Taro: I’m curious about that constraint approach part since I worry about how fast things settle; does it actually converge quickly when constraints become active?
Rosa: It does, and they found that the constraint approach achieves nearly immediate convergence, whereas the cost approach can show extremely slow convergence when those constraints kick in.
Dev: So one of the key takeaways from this paper is that the constraint approach is superior because it handles feasibility better and results in less grid loss compared to what their cost approach produces.
Taro: It sounds like they’ve found a way to make sure that when the system needs help, it finds a solution quickly without wasting time getting stuck in slow convergence issues.
The paper's summary: Rosa: Moving on to what this paper actually summarizes, they describe the OFO controller dynamic system as an update of the total derivative of an objective function based on voltage measurements and power flows.
Dev: They’re using an integrator with a constant sigma to update that objective function, and then they solve a quadratic problem inside that loop to find the next set points for active and reactive power.
Taro: The paper details how they define the terms like the gradient of the objective function, which is approximated by something called H sub p and H sub v, based on how changes in power flow affect voltages.
Rosa: They calculate these terms using sensitivities like the partial derivatives of active and reactive power with respect to power flow and disturbances, and they approximate them as constant vectors or matrices for simplicity.
Dev: The whole point of this setup is that the OFO controller iteratively changes those set points until it reaches a solution for that underlying optimization problem, which is what drives the control loop.
Taro: It sounds like they are using these gradient approximations to guide the system toward an optimal state without needing a full, complex model for every tiny change in the grid.
The paper's improvements: Rosa: Now for what they suggest as improvements, they focus on how this method can be made more robust and better suited for real-world operation. They show that the tuned OFO controller is compatible with other control devices already present in the grid.
Dev: They tested it on a distribution grid laboratory at RWTH Aachen University to prove its behavior in a real grid, and they showed that this tuned controller can deliver the requested flexibility fast while still guaranteeing operational limits.
Taro: What I like is that they show it’s robust to model mismatch and changing consumption or production in the distribution grid, meaning it uses feedback from measurements instead of just relying on static models.
Rosa: And they also tested its interaction with other devices, like an inverter doing voltage control according to VDE4105, and the controller successfully mitigates small voltage violations within one sampling interval when talking to an independent photovoltaic inverter.
Dev: They also showed that the constraint approach signals infeasibility if the requested flexibility just can't be provided by the grid, which is a necessary check for safety.
Conclusion: Rosa: So, wrapping up on "Tuning and Testing an Online Feedback Optimization Controller to Provide Curative Distribution Grid Flexibility," the paper shows that OFO with the constraint approach is a powerful way to disaggregate flexibility requests onto FPUs in a distribution grid.
Dev: It provides the requested flexibility and it signals impossibility if the grid genuinely cannot provide it, which is a solid safety feature. The tuning makes sure it’s robust to changes in consumption and production while using all of the available flexibility.
Taro: I think the main implication for me is that having a system that can handle those real-world disturbances without needing perfect models is crucial for any autonomous system operating in a grid environment.
Rosa: And it's also compatible with other control devices, meaning you can actually plug this into an existing infrastructure and get fast delivery of flexibility while respecting all operational limits.
Dev: It’s a solid tool for managing that distributed flexibility, and the constraint approach is the better way to go based on their testing results.
Taro: We're looking forward to seeing how this kind of feedback-based optimization gets applied when you try to scale this up to much larger networks or more complex scenarios.
Lukas Ortmann, Saverio Bolognani, Florian Dorfler, Fabian Bohm, Florian Klein-Helmkamp, Andreas Ulbig
Automatic Control Laboratory at ETH Zurich, Switzerland · IAEW at RWTH Aachen
eess.SY, cs.SY
Submitted: 2024-03-04
Updated: 2024-03-04
Journal ref: Electric Power Systems Research, 2024 234, 110660
DOI: 10.1016/j.epsr.2024.110660
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 90/100
The gist: Online Feedback Optimization (OFO) is presented as a new control method that steers power systems to an optimal solution of an optimization problem using minimal model information and computation…
Key concepts
- Online Feedback Optimization (OFO)
- A control method that uses real-time feedback from the grid measurements to solve an optimization problem quickly. It aims to find the best way to allocate power flexibility requests among available units using only minimal model data, making it computationally efficient for real-time use.
- Constraint Approach vs. Cost Approach
- Two ways to define the mathematical optimization problem for finding FPU setpoints. The constraint approach was found superior because it leads to lower grid losses and is more versatile, allowing the cost function to be tuned for desired system behavior, unlike the cost approach which converges slowly when constraints are active.
- Projected Gradient Descent (PGD)
- The underlying mathematical structure of the OFO controller. It uses an integral controller combined with a Quadratic Programming (QP) problem. The tuning matrix G adjusts the convergence speed of this process, where smaller values in G accelerate convergence by weighting specific variables more heavily.
Terminology
Summary
Online Feedback Optimization (OFO) is presented as a new control method that steers power systems to an optimal solution of an optimization problem using minimal model information and computation power. p. 0 One potential way to do
this is Online Feedback Optimization (OFO). OFO is a new
control method that steers power systems to the optimal solution
of an optimization problem using minimal model information
and computation power. Overall, the paper makes a clear recommendation on how to
phrase the optimization problem and tune the OFO controller. Furthermore, it experimentally verifies that an OFO controller
is a powerful tool to disaggregate flexibility requests onto FPUs
while satisfying operational constraints inside the flexibility
providing distribution grid.Where the paper says it
Where the paper says it
Problem Definition and Methodology
The core task involves defining an optimization problem whose solution provides setpoints to the Flexibility Providing Units (FPUs). The authors compare two fundamentally different ways to define this optimization problem and give a clear recommendation on which formulation is superior. p. 0 A
second challenge is to tune the algorithm that solves the optimization problem, may it be a model-based approach or OFO. To address the first challenge, we compare two fundamentally
different ways to define an optimization problem in this
context and give a clear recommendation of which formulation
is the superior one. Using a real distribution grid
setup we showed that tuning the matrix G of the OFO controller is essential for fast convergence and good performance
and we explained how to tune G in detail. We compared two
possible approaches to phrase the disaggregation task as an
optimization problem and can conclude that the constraint
approach outperforms the cost approach.Where the paper says it
Where the paper says it
Controller Structure and Tuning
The OFO controller is based on projected gradient descent, consisting of an integral controller and an underlying quadratic problem (QP). The tuning matrix G is used to adjust the convergence behavior of the QP inside the OFO controller. p. 2 Note that, G only affects the solution
of the underlying QP inside the OFO controller but does not
affect the solution of the overarching optimization problem
[28]. In other words: The overarching optimization problem
defines how the system is supposed to operate and G can
be used to adjust how the OFO controller converges to this This becomes more apparent when we multiply out
the cost function of the QP:
∥w+G
−1∇uΦ(u)∥
2
G
= w
T Gw + 2w
T ∇uΦ(u) + ∇uΦ(u)
T G
−1∇uΦ(u)
We can see that the tuning matrix G is a tuning weight in
the term w
T Gw. Therefore, small values of G lead to a small
weight on the corresponding entry in w, and the QP can make
these entries in w large while still keeping the cost function of
the QP small.Where the paper says it
Where the paper says it
Comparative Analysis of Approaches
The constraint approach is concluded to outperform the cost approach because it leads to lower losses in the grid and is more versatile as its cost function can be used to parameterize favored behavior. The constraint approach achieves nearly immediate convergence, whereas the cost approach shows extremely slow convergence when constraints become active. p. 4 Finally, we also add a charging electric vehicle which enables
us to test our OFO controller on a real grid with real loads. Performance of the Constraint Approach
In Figure 6 it can be seen that the convergence of the
constraint approach is nearly immediate.Where the paper says it
Experimental Validation and Robustness
The OFO controller was tested on a real distribution grid laboratory at RWTH Aachen University to show its behavior in a real grid. The results demonstrate that the tuned OFO controller is compatible with other control devices and enables fast delivery of the requested flexibility while guaranteeing operational limits. p. 6 provide the requested flexibility and can signal this to the
entity requesting the flexibility. Furthermore, the experiments
showed that the tuned OFO controller is compatible with
other control devices in the grid and enables fast delivery
of the requested flexibility while guaranteeing the operational
limits in the distribution grid. Furthermore, the experiments
showed that the tuned OFO controller is compatible with
other control devices in the grid and enables fast delivery
of the requested flexibility while guaranteeing the operational
limits in the distribution grid. Also, the method is robust to
model mismatch and changing consumption in the distribution
grid while it is able to utilize the full flexibility of the
distribution grid. Control performance of the constraint approach. grid is providing the requested flexibility while operating the
distribution grid within its operational limits.Where the paper says it
Where the paper says it
Where the paper says it
Interaction with Grid Components
The OFO controller interacts well with other devices in the grid, such as an inverter performing voltage control according to VDE4105. The controller mitigates small voltage violations within one sampling interval when interacting with an independent PV inverter. p. 5 The dashed orange line in Figure 7 shows that deviations
from the setpoint are immediately mitigated in the next time
step, meaning once the controller detects a deviation it fixes it
within one sampling interval and accurately. Figure 7 shows
that small voltage violations can occur when the independent
PV inverter II changes its active power injections. Within a few iterations,
p and q converge to the optimal solution of the overarching
optimization problem. OPTIMIZATION PROBLEMS FOR DISTRIBUTION GRID
FLEXIBILITY
The FPU are supposed to deliver the requested flexibility
Pset at the PCC while all voltages v, active power injections
p, and reactive power injections q are within their limits.Where the paper says it
Where the paper says it
Conclusion on OFO Utility
Overall, OFO with the constraint approach is a powerful tool to disaggregate flexibility requests onto FPUs in a distribution grid. The method provides the requested flexibility and signals impossibility if the grid cannot provide it. p. 6 An electric vehicle is charged and the OFO controller counteracts this
disturbance such that the distribution grid accurately provides the requested
flexibility while satisfying the distribution grid constraints. provide the requested flexibility and can signal this to the
entity requesting the flexibility. Furthermore, the experiments
showed that the tuned OFO controller is compatible with
other control devices in the grid and enables fast delivery
of the requested flexibility while guaranteeing the operational
limits in the distribution grid. Also, the method is robust to
model mismatch and changing consumption in the distribution
grid while it is able to utilize the full flexibility of the
distribution grid. Furthermore, the experiments
showed that the tuned OFO controller is compatible with
other control devices in the grid and enables fast delivery
of the requested flexibility while guaranteeing the operational
limits in the distribution grid. Also, the method is robust to
model mismatch and changing consumption in the distribution
grid while it is able to utilize the full flexibility of the
distribution grid.Where the paper says it
Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improved AI system can perform fast disaggregation of flexibility requests onto Flexible Providing Units (FPUs) by using an Online Feedback Optimization controller that steers power systems to an optimal solution of an optimization problem using minimal model information and computation power.
-
The improved system can be made robust to model mismatch and changing consumption and production in the distribution grid because the Online Feedback Optimization method incorporates feedback from the grid through measurements, as opposed to relying solely on offline models.
-
The AI system can accurately and quickly disaggregate a curative action request onto a heterogeneous group of FPUs in a distribution grid while guaranteeing the satisfaction of operational constraints inside that flexibility providing distribution grid.
-
The system can be tuned for fast convergence and good performance by tuning the matrix G, which is essential because
G only affects the solution of the underlying QP inside the OFO controller but does not affect the solution of the overarching optimization problem.
-
The AI system can outperform a cost approach by using reactive power as part of its cost function, leading to lower losses in the grid and more efficient utilization of inverters.
Abstract
Due to more volatile generation, flexibility will become more important in transmission grids. One potential source of this flexibility can be distribution grids. A flexibility request from the transmission grid to a distribution grid then needs to be split up onto the different flexibility providing units (FPU) in the distribution grid. One potential way to do this is Online Feedback Optimization (OFO). OFO is a new control method that steers power systems to the optimal solution of an optimization problem using minimal model information and computation power. This paper will show how to choose the optimization problem and how to tune the OFO controller. Afterward, we test the resulting controller on a real distribution grid laboratory and show its performance, its interaction with other controllers in the grid, and how it copes with disturbances. Overall, the paper makes a clear recommendation on how to phrase the optimization problem and tune the OFO controller. Furthermore, it experimentally verifies that an OFO controller is a powerful tool to disaggregate flexibility requests onto FPUs while satisfying operational constraints inside the flexibility providing distribution grid.
Sources
Related papers
- Industrial Load Modeling and Optimization for Market-Based Interaction with Power Systems
- Exact and Approximate Convex Reformulation of Linear Stochastic Optimal Control with Chance Constraints
- Certified Set Convergence for Piecewise Affine Systems
- Symbolic Control for Stochastic Systems via Finite Parity Games
- Multiobjective optimization-based design and dispatch of islanded, hybrid microgrids for remote, off-grid communities in sub-Saharan Africa
- Physiology-informed layered sensing for intelligent human-exoskeleton interaction