Locally Distributed Real-time Co-simulation Infrastructure for Scala ble System Simulation (LD-RTCosim)

Power system analysis requires real-time simulation via Digital Real-Time Simulator (DRTS) to reproduce the complex behaviour of transmission and distribution grids with high diffusion of Distributed Energy Resources (DER) with the aim of enhancing the process of energy system integration~\cite{europeancommission_communication_2020}. These systems not only permit solving power system network by means of nodal analysis but also accelerate the study and development of power system equipment by means of Hardware-In-the-Loop (HIL) and Power Hardware-In-the-Loop (PHIL) application. To this extent, different DRTS commercial solutions (e.g., OPAL-RT and RTDS Technologies) have been proposed to accelerate the required computation for highly intensive tasks, such as Electro-Magnetic Transient (EMT) analysis~\cite{babikir2021acompensated}. However, the real-time simulation of innovative smart grids calls for huge computation capabilities, mostly to coordinate complex systems for largescale scenarios. 

Different works in the literature already implemented parallel computing strategies to solve EMT analysis of scalable Power System Under Test (PSUT). For instance, the MATLAB Simulink Simscape library~\cite{simscape} allows splitting a power system by using the propagation delay of a transmission line to absorb the inherent delay required to avoid direct feedthrough. Classic power systems present many transmission lines long enough to decouple points in the network, allowing parallel computation on multiple cores of a Central Processing Unit (CPU)~\cite{dufour2015testing}. The same concept is applied in ARTEMiS library~\cite{artemis} to split a power system on different cores of OPAL-RT Technologies DRTS, enhancing their real-time scalability~\cite{faruque2015realtime}. However, the computational capability of a single DRTS is limited and cannot overcome the problem of analysing large smart grid scenarios.  

Thus, the scientific community proposed to interconnect different DRTS racks by means of communication protocols (e.g., Ethernet) to join their computational capabilities in a unique distributed digital real-time power system co-simulation environment~\cite{arjen2017cyberphysical}. Communication latencies, synchronization, and time regulation are the main issues in these interconnections, where the time step duration reach tens of microseconds. So, to accomplish this target is necessary to develop new technologies that allow DRTS systems to be connected without compromising simulation performance, particularly in terms of the accuracy of the dynamic simulation. Most of the proposed solutions in the literature are inspired by the PHIL application and its stability and accuracy intrinsic issues~\cite{zhiwang2020ascheme}. For instance, the time delay compensation method presented in~\cite{guillo2021characterization}. However, this strategy restricts the analysis to slower dynamics than EMT ones. Other works are inspired by the Interface Algorithm (IA) proposed for PHIL real-time simulation. The most suitable IA for these interconnections is the Ideal Transformer Method (ITM)~\cite{barbierato2023sj}. However, the effect of the latency experienced by the signal exchanged (i.e. voltages and currents) still generates a noticeable non-linear effect, affecting the accuracy of the real-time co-simulation solution. As a matter of fact, none of the previous solutions permits linking with flexibility different DRTS together maintaining the accuracy of a monolithic simulation. 

In this LD-RT-Cosim project, we developed a digital real-time power system co-simulation implementing a Distributed Transmission Line Model (DTLM) as an adapter to split and extend the scalability of a PSUT with respect to a monolithic digital real-time simulation. The DTLM is applicable to transmission lines with a minimum length determined by the characteristic propagation velocity 𝜈 of the line. The co-simulation setup is accomplished by interconnecting two DRTS by means of an optical fiber link between their Small Form Pluggable (SFP) ports, exploiting the Aurora 8B/10B protocol. Aurora, in a nutshell, ensures the minimum communication latency among the available protocols on commercial DRTS. The proposed solution absorbs the communication latency experienced during the data exchange management of the co-simulation setup in the propagation time of the travelling wave of the DTLM. Because the communication latency is typically a multiple of the time step duration TS depending on the commercial DRTS setup~\cite{barbierato2023sj}, the proposed DTLM allows for the setting of a variable communication latency attribute based on the laboratory configuration. To avoid complex time regulation and synchronisation schemes, the co-simulation setup has been tested over a laboratory setup with a single  DRTS by employing an optical fiber echo link among two SFP ports of an OPALRT OP5700. The proposed solution has been tested over a simple PSUT with the aim of 
demonstrating the stability and accuracy of the co-simulated experimental result, resulting in identical performances with respect to a monolithic simulation.