Dynamic Link Library (DLL) applications are a powerful approach for modelling of as-built functionality of the vendor-specific equipment and control systems to be shared with the clients without divulging strictly privileged information, intellectual property or know-how of the vendors.

The clear advantage of the approach is that the vendor can convert complete and sufficient models of the equipment and control systems from a preferred by the vendor simulation software, such as C++, into the DLL applications to be used with the client’s simulation software, such as the DIgSILENT PowerFactory®, for dynamic stability assessment of the electric power grids. Though the electro-magnetic transient (EMT) modelling can be included into the DLL applications and used for the RMS assessment of large power grids including symmetrical and asymmetrical short-circuits and island operations with a passive network without compromising the execution time.

Among pronounced challenges of the DLL applications can be: (i) – difficulty to perform fully variable time-step simulations, i.e., automatic adaption of the time step within 1 to 10 ms, (ii) – appearance of numerical discontinuities about the simulation events, i.e., so-called numerical spikes having noticeable magnitudes in voltage and reactive power at abrupt voltage changes following application or clearance of short-circuits. The numerical spikes are not present in measurements. Furthermore, the numerical spikes may affect the own control models and propagate to the control models of other equipment models distorting the system-level simulation results.

This presentation describes a method developed by Siemens Gamesa Renewable Energy (SGRE) for overcoming the mentioned challenges of the DLL applications. Using simple and understandable validation cases, the developed method will be demonstrated for an EMT level modelling of a generator in the RMS simulations using the DIgSILENT PowerFactory® software.

The developed method enables numerically robust application of fully variable time-step simulations, such as within a commonly required range from 1 to 10 ms. The developed method efficiently suppresses numerical spikes that would appear at abruptly voltage changes, such as after clearance of short-circuits and allows simulations in an island regime with a passive network.