Learnings

Outcomes

The main outcomes for the project are outlined below. For convenience, they have been split into outcomes from each respective Work Package.
Work Package 1 Outcomes
· Literature review – This literature review was delivered as a report and addressed how distributionconnected inverters associated with renewable energy resources may provide harmonic compensation as an ancillary service. Investigations were made and analysis was undertaken on the importance of harmonics and the different type of filters that are used to reduce them.
· Model Development – The model development compromised of the Work Package 1 report3. This report documented the creation and validation of the modelling environment within MATLAB/Simulink. An in depth analysis of the stages taken to authorise the model was given including all electrical characteristics.

Work Package 2 Outcomes
· Work Package 2 report – This report4 documented the development of the control algorithm that would be utilised on a single inverter to provide harmonic compensation. Information captured in this report centres around the choices of control parameters used to generate the correct level of harmonic compensation.
· Work Package 2 slide pack – This slide pack accompanies the Work Package 2 report5 and expresses the Work Package 3 Outcomes
· Work Package 3 report – The Work Package 3 report6 further developed the work carried out in the previous work package. This included adding the AF functionality to multiple inverters rather than a single inverter unit within the network model. It was also detailed within the report the improvements made to the algorithm compared to the designed one in Work Package 2.
· ISGT Europe 2021 Paper – “Harmonic Mitigation as Ancillary Service Provided by Multiple Photovoltaic Inverters”. This academic paper described the work carried out up until Work Package 3. Presented at the Innovative Smart Grid Technologies Europe 2021 conference and published on the IEEE online library, this paper detailed the specific control theory that was behind the algorithm and showed its effectiveness in reducing network harmonics. The paper can be purchased and read here: Harmonic Mitigation as Ancillary
Service Provided by Multiple Photovoltaic Inverters | IEEE Conference Publication | IEEE Xplore
· CIGRE 2022 Kyoto Symposium, Japan – “Use of PV inverters to mitigate harmonic levels on distribution systems” Similar to the ISGT paper, this symposium report included an updated control algorithm with additional results due to the further supplements added to the control algorithm. Presented April 2022.
· ENIC 2021 Poster – this poster was presented at the Energy Networks Innovation Conference 2021 where the project was exposed to a high number of stakeholders across industry.
Work Package 4 Outcomes
· Work Package 4 report – The Work Package 4 report captured the analysis undertaken to integrate the control algorithm into a physical inverter and test it on a network model through real time simulation. This included the encountered problems, the steps taken to resolve & mitigate them and a comparison exercise of results to those seen in Work Package 2.
· RTLab21 Presentation – A presentation was given at the OPAL-RT RT21 conference. The presentation had a focus on how the real time simulation element of the Work Package functioned and the steps involved to incorporate the algorithm into the system and run successful simulation.
Post Project Activities
· Technical Summary of Harmonic Mitigation report7 – This work was completed by PSC and provided a high-level summary of all the work undertaken throughout the duration of the project.
· WPD Closedown report8 – This work was completed by WPD and provided an overview of the projects’ activities through a project management lens.
The full learning has been disseminated via WPDs Innovation website page where access to full work package reports and both closedown reports is freely available at: www.westernpower.co.uk/Harmonic-Mitigation

Lessons Learnt

During the execution of the project, several difficulties were encountered and these were resolved using the bestengineering approach possible.

Below is a list of key takeaways that could have been anticipated at earlier stages:
1.1.1 Understanding of analysis platform’s limitations at the onset of the project
The suitability either of PSCAD, RSCAD and RTDS, stand-alone or in combination with OPAL-RT 6500 and/or MATLAB/Simulink could have been examined to find the best alternative.
The limitations experienced were as follows;

· Lack of tap changer in transformer models in MATLAB/Simulink.
· Lack of frequency dependency modelling for most of the power network elements in MATLAB/Simulink.
· Inadequacy of the emulator OPAL-RT 6500 in accommodating small integration time-steps.
· Capability of the emulator OPAL-RT 6500 to accommodate only a very small network model.
· Observation of numerical stability issues with the time domain solvers in both MATLAB/Simulink and OPAL-RT
6500.
· The need to introduce a ‘magic’ resistor to the model both in MATLAB/Simulink and HIL simulations (howeverit should be noted that the effect of this is negligible on study results and is a common approach employed innumerical integration studies).
1.1.2 Knowledge of time samplings on measured data needs to be known or planned in advance ofstarting the project execution.
· The input data utilised for models based on SCADA and temporary Power Quality (PQ) monitors were observed to be in significantly different time resolution.
· The time resolution of measured data was not suitable for the discrete time-steps in the real-time simulator.
· The observation of the overall simulation was computationally intensive. This type of difficulty would be seen in other electromagnetic transient analysis environments also with the main driver being the time step and the overall duration of the simulation time. The largest time step that can be used is dictated by the model characteristics and not much can be done about it. The duration of the simulation is split into multiple parts so
that simulation time is shortened, and the data generated can be handled easily.
1.1.3 Control algorithm improvements that improved the performance
· Replacement of the low pass filter with a better performing notch filter to avoid oscillations in harmonic reference signal.
· Introduction of an automatic gain such that the inverter is not overloaded under high irradiance conditions.
· Separation of gain from applying to a single harmonic to a grouping of harmonics thus providing balanced compensation between harmonic orders.
· Introduction of a Proportional Resonant (PR) controller instead of a Proportional Integral (PI) controller while generating harmonics for injection.
· Introduction of additional filtering to cater for unbalanced conditions in the harmonics.
A final point to make regarding learnings relate to observations during the execution of the project.
· Whilst an optimum location of feedback measurement was established for normal operation, it is possible to have multiple measurement points with the possibility to modify them, depending on system operating conditions or other events (for example, faults on a feeder).
· The inverters have an impact on the equivalent network impedance, in particular at the frequencies of current injection, due to a combination of inverter output filter and control loops.