By Francis Luces, Ric Austria

Power projects planning to participate in the wholesale market are required to undergo impact studies as part of the interconnection application process. The studies, as a minimum, evaluate the performance of the projects under instantaneous, steady-state and transient conditions. The timescales of phenomena and equipment studied are as illustrated in Figure 1.

In the specific case of transient studies, the impact of a proposed project on the voltage and frequency control capability of the overall grid is evaluated. Traditionally, it was sufficient to consider a timeframe of 0.5 to 10 Hz (10-100 msec) for a type of study known as transient stability. The computer models (and software, such as PSS/E and PSLF) used to conduct these studies are known as phasor-based models. These models capture phenomena limited to the target timeframe.

The recent rise in interconnections for inverter-based energy resources, whose control response reach into the microsecond (usec) timeframe, has brought about renewed concern for phenomena such as switching transients, lightning impacts, fault response, transient and temporary overvoltage, among others. This is due to the very fast response time of inverter control schemes. The expanded timeframe for analysis now reaches into the electromagnetic transient (EMT) regime requiring different models and software (a common software in use in the US is PSCAD/EMTDC). Utilities and system operators overseeing interconnection processes are now requiring that EMT models be provided as part of the application.

Furthermore, since there is some overlap in the time regime of phasor-based and EMT models, there is a need to benchmark the dynamic response to ensure that the models (built for disparate software such as PSS/E and PSCAD) behave similarly. This benchmarking is also now being required by entities such as ISO New England and the Hawaiian Electric Company.

 

Modeling and Simulation Considerations

 

Typically, phasor-based models of conventional generators and their associated power plant control systems are studied using a power system dynamic simulation (PSDS) program to simulate generator dynamics and control action in the range of 10-100 microseconds. For inverter-based generation such as solar PV, Type III and Type IV wind farms, and energy storage systems, the range of response is broader with a higher frequency of interest. The representation of these technologies in dynamic simulation is modeled in a practical way where some fast dynamics in power conversion applications are simplified to obtain the essential response. To get the exact dynamic characteristics associated with the complexity of models of inverter-based generation, the need to use electromagnetic transient programs (EMTP) becomes apparent. The main characteristic of EMTP is that it allows numerical integration of differential equations having very small time constants, capability in performing three-phase simulations, and simulation of very fast transient phenomena. Also, it is capable of modeling control systems, thus allowing inner control loops of inverter-based generation to be modeled in detail. Therefore, any control action and power system event in the microsecond range and within the timeframe of stability phenomena could be studied in the benchmarking process. Once the responses of two different models are verified, these can now be utilized for the conduct of more advanced studies.

 

A Benchmarking Example

As an example, three models of inverter-based energy storage project are considered for the purpose of benchmarking: (1) a generic model available in PSDS dynamic model library with manufacturer-specific parameters, and two user-developed models (one for PSDS and another for EMTP). Basic tests such as initialization, flat run, and ringdown simulations are carried out to determine the response of each model. To compare the responses, power system parameters such as active and reactive power output, and voltage and frequency measured at the project’s point of interconnection (POI) are obtained by simulation. The responses of each monitored parameter are overlaid as shown in Figures 2 through 5 below.

 

 

In the benchmarking process, three regions in the plot are of key interest:

1. Initialization (Pre-Fault) period – at time < 1 second, the response of EMT model do not match the values from the dynamic simulation of generic and user-written models. This is expected since dynamic simulation initializes from the power flow solution of the network while the EMT simulation calculates first the instantaneous values of voltage and currents at the terminals of the inverter. However, each of the models simulated should achieve a steady-state or quiescent point of operation which in this case reached at time = 2 seconds.
2. Fault period – one of the requirements in benchmarking is the disturbance test where typically a three-phase fault cleared by circuit breakers is applied at the project’s POI for verification of project’s responses. In the plots shown above, the response of each model follows a common trend, but spurious high-frequency spikes were present. The spikes may be due to the model’s behavior at the instant of fault or during fault recovery which sometimes leads to non-convergence of simulation. It may be also attributed to how PSDS or EMT software treat an abrupt change (or a discontinuity) in the response.
3. Post-Fault Period – after the fault is cleared, the post-fault response of each model should match with one another and should achieve their pre-fault values or gain a new operating point. In this case, the steady-state response of each model was achieved at time = 8 seconds after clearing of the fault.

In addition to the basic tests, some utilities also require benchmarking of model responses for inverters with smart functions. For example, step tests are performed to evaluate reactive power (volt-var) and active power (frequency watt) responses when variations in voltage and frequency are applied at the POI. An example of voltage step test comparing responses from EMT- and phasor-based inverter models having default controller parameters is shown in Figure 6.

 

 

In this test, variations in voltage at the POI are applied with a specific duration without triggering the protection settings of the inverter. It is desired that for every change in the POI voltage, there should be a corresponding change or response in the reactive power output of the inverter. The reactive power output across the models are compared. In the figure above, the parameters of the EMT model needs to be modified to match the response of the phasor models.

 

Takeaways

The acceptability of models depends on how good a fit the responses show. It is important to note that since EMT models capture higher frequency phenomena that EMT response may show additional fluctuations or peaks and dips than the phasor model. In benchmarking, several modeling concerns can arise:

• Scalability of models – some transient models of inverter-based DG do not have capability to scale the capacities i.e. changing the MW capacity for a specific application. Creating multiple instances of the transient model to scale the capacity may be a good approach to work on this limitation but this a tedious task especially for large projects having hundreds of inverter units. This modeling approach may cause slower simulation compared to a scaled model.
• Special model capabilities (smart inverter functions) – inverter manufacturers may introduce control modes and smart inverter functions. The specific features applied to the simulation needs to be identified since they affect compliance to the standards of interconnecting utility and may impact the bulk power system performance.
• Model documentation – the completeness of user-developed model description, capabilities, and potential limitations is very helpful for the users of the model (e.g. planners, operators, and plant owners). The model documentation is very important to understand the user-developed model’s capability and is critical during the benchmarking process.

As more inverter-based generation is integrated to the electric power system, a higher demand for time and effort is required to look into these specialized models using key power system analysis tools described above. In addition, building the capability and skills of power system engineers doing the benchmarking analysis is also equally important to respond to modeling and simulation needs of bulk power system with inverter-based generation.