With an increasing share of local generation on European grids, new grid codes have been defined to ensure secure network operation. Within these grid codes, simulation models are often required to prove the compliance of a generation unit with the requirements of the codes. Eckehard Tröster looks at how these models are validated in Germany with regard to CHP installations.

 

New European grid codes describe how local generators must act during normal and critical situations.
New European grid codes describe how local generators must act during normal and critical situations.
Credit: National Grid

With an increasing share of electricity production from renewables and combined heat and power (CHP) plants, grid operation must be revised as more and more conventional power plants, which used to provide ancillary services for grid operation and system stability, are taken offline. For this reason, new grid codes have been set up describing the requirements for local generation units on how to act during normal and critical grid situations.

For economic and technical reasons, proof of compliance with these requirements cannot always be demonstrated by measurements. Thus, simulation models have become an important part of the confirmation procedure. Here we will examine how simulation models are used in Germany to prove compliance by validation and certification. The main focus is on models of CHP plants with synchronous generators, which is currently one of the hottest topics in the German modeling, validation and certification scene.

Grid code requirements

Grid codes may set a large number of rules on the operation of assets within the power system, not all of which are relevant to local generation. If very few local generators feed into the grid, they can get away with whatever behaviour is cheap and easy to implement, as their influence on grid stability can be neglected. However, as the share of local generators rises, they must take over a growing number of duties from the conventional generators they replace.

The role of grid code requirements is to ensure that local generators acquire the technical capabilities to take over these duties. Grid codes are necessary for all voltage levels, with system stabilisation requirements that were previously only demanded from conventional generators in the high-voltage grids. Requirements set out in grid codes for local generators therefore range from basic power quality requirements and system support in normal operation to specific behaviour during grid disturbances, such as short circuit events in the transmission system.

Historically, local generators were required to disconnect from the grid in case of a short circuit, a behaviour which is easy to implement, but only acceptable at low shares of local generators. New grid code rules require local generation to stay connected during a fault (low voltage ride through, or LVRT).

Theses requirements go back to studies conducted between 2003 and 2005, which showed severe problems in northern Germany in simulated fault cases. Disconnection of wind turbines during grid faults would, in some cases, have led to the loss of more generation than the European primary operating reserve could handle, and therefore would have increased the risk of a blackout of the European power system.

LVRT and CHP

As small gensets with synchronous generators typically have a comparably small inertia, the generator will accelerate fairly fast if the electric torque is reduced because of a voltage dip in the power grid.

As already noted, many grid codes require the genset to stay connected in case of a voltage dip in the grid for a certain time (typically in the range of 150 ms to 1.5 seconds depending on the voltage level). As the genset starts to accelerate, it might either trip due to overspeed protection or – more likely – suffer a pole-slip and therefore lose synchronisation to the grid frequency. Energynautics was commissioned by the VDMA (the German engineering association) to investigate possible solutions to this situation by means of modeling and simulation.

An obvious solution would be to reduce the mechanical torque of the engine. However, as there is typically some fuel left in the engine and the time-constant to reduce the fuel is larger than the time the voltage dip lasts, this is not a suitable solution in most cases. Instead, many other different solutions have been investigated, such as additional resistors, use of a full-power converter, adding a flywheel and use of an eddy-current brake. Some were more promising than others; however, all of these solutions were fairly expensive as either additional equipment would have had to be installed or the construction of the CHP facility would have had to be revised.

Figure 1. LVRT requirements for synchronous generators, according to Germany's medium-voltage grid code
Figure 1. LVRT requirements for synchronous generators, according to Germany’s medium-voltage grid code

With the help of simulation results, we were able to show that a minor adjustment of the German medium-voltage grid code would actually make it possible for the majority of the existing CHP installations to ride through the fault without slipping. This adjustment is related to the voltage level during the fault. For synchronous generators connected directly to the medium-voltage grid, the requirement has been changed so that below 30% residual voltage, the CHP may disconnect from the grid (see Figure 1). Originally 0% had been requested; however, as the synchronous generator lifts the voltage level in the case of a short circuit in the grid by providing reactive current, a voltage level of 0% will never be seen at the terminals, at least as long as it is not a bolted short circuit, which in turn would not be relevant for system stability.

Model validation in Germany

Many critical situations in the grid cannot be tested in a real environment as this would lead to a high risk of blackouts. Therefore, grid operators often use computer simulations to predict the behaviour of their grid, especially the transient behaviour in fault cases or extreme situations. In order to get reliable and representative results from their model, the operator has to know the properties and behaviour of the generators on their grid. Generator owners are thus often required to hand in simulation models in a specified format before they are allowed to connect their unit to the grid. The usage of models and simulation is therefore an important part of the German certification process to prove grid code compliance.

In order to prove the quality of the model, it must be validated against measurement data. For the certification process in Germany, LVRT measurements must be carried out. These measurements are described in the technical regulation No. 3 by the FGW (the federation of German wind power and other renewable energies). For CHP plants, 14 different tests must be carried out under different conditions, such as voltage levels, reactive power, type of fault and loading level of the CHP system. These tests are conducted by a LVRT-container, which emulates a voltage dip at the terminals of the test unit without a major impact on the grid itself (see Figure 2).

Figure 2. One line of LVRT test setup
Figure 2. One line of LVRT test setup

In FGW Technical Regulation No. 4 (FGW TR 4) the validation procedure is described, along with how to use the derived measurements and compare them to simulation results. The main idea is to split the test sequence into three intervals: before, during and after the voltage dip. These intervals are again each split into a dynamic and a steady-state period. For each period an average error is determined by comparing the average value of the measured signal with the average value of the simulation. Additionally, for the steady-state intervals, the maximum instantaneous error is determined by comparing the measured and simulated signal at every instant in time. This procedure is carried out for active power, reactive power and reactive current.

Depending on the type of signal and investigated interval, errors in the range of up to 10%-20% are allowed in order to judge the model as accurate enough and therefore validated. Additionally, the load angle is determined from speed measurements and compared to the simulation result.

Figure 3. Model structure of IEEE Type AC8B - alternator-rectifier excitation system
Figure 3. Model structure of IEEE Type AC8B – alternator-rectifier excitation system

For CHP facilities there are two major reasons for providing a validated model:

1. The validated model is used to prove the fault ride-through capability of other gensets that have not been measured but are technically very similar;

2. In addition, simulation models are used to determine the dynamic behaviour of specific projects under real grid conditions.

CHP modelling

The most relevant parts of the CHP plant regarding impact on genset behaviour during and after a voltage dip are the generator, the exciter, the automatic voltage controller (AVR) and the reactive power controller. Although these systems are often very similar, there are several different approaches to provide excitation and control voltage and reactive power.

The most common systems for the power supply of the exciter machine are auxiliary windings and permanent magnet generators. Auxiliary windings are fairly cheap but can influence the output of the AVR during the fault considerably as during the fault the AVR typically opens up completely in an effort to provide as much excitation as possible – which again is limited through the power supply.

For the AVR and reactive power controller, the most common approach is still to have both active and then send an offset signal from the reactive power controller to the voltage controller, which then controls the excitation system. This configuration has the advantage that the excitation is raised automatically during the fault, and therefore stabilises the generator.

With the development of digital controllers a new control system has been introduced, where the reactive power controller acts directly on the excitation system. In this case, the power factor controller tries to reduce the excitation during the voltage dip as a high reactive power is supplied to the grid. This in turn destabilises the genset to a certain extent, although in the cases we have investigated, the gensets still rode through the fault without becoming unstable.

Due to this drawback, another concept is to switch – in the case of a voltage dip – from reactive power control to voltage control until the voltage is back. This concept makes it possible to increase the excitation during the fault, and therefore further stabilises the genset.

Concerning the reactive power controller, again different methods exist with and without a detection of the voltage dip leading to a freezing of the controller or even making it faster.

Due to the variety of possible control structures, it is very difficult to develop a common model that is valid for all manufacturers as was originally intended. Instead, manufacturer-specific models have been developed.

Europe's new grid codes require validation through simulation models
Europe’s new grid codes require validation through simulation models
Credit: European Union

German certification

According to the German medium-voltage grid code and its supplementary documents, for CHP plants the capability of LVRT has to be proven by certification since the beginning of 2014. A temporary regulation was then introduced, which allowed the certificate to be provided by the end of that year. As a large majority of manufacturers was not able to deliver the required certificate, also because of the lack of simulation models, the BDEW (German association of energy and water industries) announced a recommendation for grid operators to extend the deadline for certificates until the end of June 2015.

As the previously described validation procedure had been developed for models of wind and solar power plants, it turned out that, in many cases, it is not well suited for synchronous generators. Just to name two of these issues:

1. The required accuracy of the model is defined in absolute values related to the nominal current of the generation unit. As for synchronous generators the short circuit current is typically a multiple of the nominal value, the requirement for accuracy in relation to the actual value is very tough. This issue has not been observed for wind or solar power, as the converter-based generation units can typically only provide currents up to the nominal value;

2. Due to the fact that CHP facilities with synchronous generators are directly connected to the grid and the exciter machine with its time constant and power supply, we do observe active power oscillations during and after an LVRT event. Additionally, the dynamics of the reactive power typically last much longer compared to converter-based generation. This makes it much more difficult to develop models reaching the required accuracy according to FGW TR 4.

For these reasons a task force was installed to discuss a different procedure, taking the properties of CHP plants into account. One possible solution is to use a validation procedure as it is described in IEC 61400-27-1 for wind turbines. Here the required accuracy limits are not defined, and therefore can be set according to the achievable accuracy from a cost-benefit perspective. However, the different intervals defined in the standard are again very much linked to the capabilities of wind models, and therefore may not be adequate for synchronous generators.

Looking forward

With an increasing share of local generation in the grid, new grid codes have been defined to ensure secure grid operation. Within these grid codes, simulation models are often required to prove the compliance of a generation unit with the requirements of the grid codes. In order to judge the reliability of the model, the developed models have to be validated. In Germany this is done through a certification process.

Although the CHP market is dominated by synchronous generators, there are a couple of differences from one manufacturer to the other. These differences do influence the dynamic behaviour of the generation unit to such an extent that the well-known IEEE models are, in most cases, not sufficient to reach the requested accuracy demanded in the FGW Technical Regulation in order to fulfill the German grid codes. For this reason it is necessary to develop manufacturer-specific models, which take the individual control strategies into account.

Besides the necessity of manufacturer-specific models, it must be noted that the described validation procedure has been developed for converter-based generation units, such as wind and solar power, and therefore the validation procedure is not well aligned to the properties of synchronous generator-based CHP systems. Necessary changes in the procedure are currently under discussion in the relevant working group. The new regulations for validation are expected to be published in the course of this year.

Dr Eckehard Tröster is Senior Engineer at Energynautics. www.energynautics.com