Model Aggregation Methodology for Power System Study

In our busy professional life, we miss to refer many a useful document to answer some of our common but vital queries. One such query which I have faced several times, is how to aggregate the generators of Windfarms and SolarPV plants for power system study.

If you also have the same query, then you may find the following extract from Australian Electricity Market Operator document useful.

Model aggregation

Traditionally there has been a one-to-one correspondence between power system elements such as generating units and the models of these elements in simulation software. Thus, each generating unit has been represented individually in the power system model. This is practical when the power system plant is a large power station comprising up to about a dozen individual large generating units. However, contemporary generating systems such as wind and photovoltaic solar farms, as well as other plant such as grid-scale battery installations, can include as many as several hundred generating units. As these generating units are usually identical to one another, this has the effect of multiplying the required computational effort and simulation run time for little benefit, compared to representing these identical generating units as a smaller number of aggregates.

For such generating systems comprising of scores of small generating units, the general rule is that the submitted plant model should contain no more than four generating units of any one type. That is, generating units should be combined into aggregates with each aggregate representing multiple individual generating units. In the simplest cases, a single aggregate may suffice to represent the totality of generating units.

Aggregation should not be used to combine power system elements of differing types. These should retain separate explicit model representations, albeit some may be aggregates of identical units of that same type. An exception may be made where elements are similar in all material respects other than size (for example a 3.0 MW and a 3.2 MW wind turbine with the same underlying technology and control systems) and where evidence is provided of this similarity by way of manufacturer documentation, to the satisfaction of the Network Service Provider (NSP) and Australian Electricity Market Operator (AEMO).

Scaling principles for derivation of multiple-unit aggregates?

The following general principles are assumed as the default for producing aggregates of N identical units, where each unit is assumed to consist of a ‘plant’ at low voltage (LV) in cascade with a unit transformer stepping up to medium voltage (MV).

The MV ‘collector system’ which connects the individual generating units together is treated separately in the next Section.

Where the modelling of power system plant requires an aggregation method that varies from these principles, this must be clearly documented in the RUG. Alternative aggregation methods include the provision of a separate aggregate model not directly derived from the individual unit model. Evidence must be submitted to AEMO and the relevant NSP for the suitability of the aggregation method relative to the simple application of the scaling principles below. AEMO and the NSP must jointly assess this evidence, and may accept the different method, or determine that the scaling principles will apply if the evidence submitted is weak.

· The aggregate generating unit is represented in the model in an analogous fashion (size aside) to a single generating unit. It has the same associated dynamic model and appears similar to a generating unit in the network model in cascade with an equivalent unit transformer.

· The LV and MV voltage levels are the same for the aggregate as for the individual generating units.

· The MVA rating of the aggregate plant is N times the MVA rating for an individual generating unit. (This rating is called MBASE in the PSS?E software.)

· The active power and reactive power of the aggregate are the sums of the individual generating unit powers. For modelling purposes, there is an underlying methodological assumption that each generating unit has identical power outputs, although these will vary from unit to unit.

· Any other ‘size quantities’ specified in SI units, or in per-unit on a fixed system MVA base, are multiplied by N in the aggregate. Examples of size quantities are rated current in Amperes, rated torque in Newton-metres, and inertia constant in Joules or VA-seconds (but not speed or voltage). Where, on the other hand, the model specifies these quantities in a per-unit system on the generating unit MVA base, their numerical values are identical.

· The MVA rating of the aggregate generating unit transformer is N times the MVA rating of each generating unit transformer.

· Any internal series impedances of the aggregate generating unit, generating unit transformer and any intervening LV cables, when specified in ohms or in per-unit on a fixed system MVA base, have values 1/N times their values for each corresponding generating unit. Where, on the other hand, the model specifies these quantities in per-unit on the unit MVA base, their numerical values are identical.

· Any internal shunt admittances of the aggregate generating unit, generating unit transformer and LV cables, when specified in Siemens or in per-unit on a fixed system MVA base, have values N times their values for each corresponding generating unit. Where, on the other hand, the model specifies these quantities in per-unit on the unit MVA base, their numerical values are identical.

Implicit in these scaling principles is a requirement that the underlying model of the unit is also capable of representing the aggregate of N units when configured with the larger MVA rating. If appropriate, the model may be coded to indicate the level of aggregation explicitly in the model configuration (for example, by including either each unit size or the number N of identical units as a configuration parameter). However, any necessary change to model configuration or settings beyond those stated above when switching between an individual unit and aggregate representation, or between aggregate representations with different numbers of units, must be clearly documented in the RUG.

Representation of collector systems in aggregated models?

Special attention must be given to the aggregated representation of the MV ‘collector system’ that connects the MV terminals of the generating unit transformers and (usually) conveys the aggregate generated or consumed power to an MV collector bus at the relevant substation.

In the simplest case, all identical generating units are combined into a single aggregate, and the model specifies a single equivalent collector impedance connected between the MV collector bus and the MV terminal of the aggregate equivalent generating unit transformer. In this case, the recommended procedure for calculating the equivalent collector impedance is given in National Renewable Energy Laboratory (NREL) report NREL/CP-500-42886, “Method of Equivalencing for a Large Wind Power Plant with Multiple Turbine Representation”34. This procedure is based on calculating the equivalent series resistance and reactance that yield the same active power and reactive power consumption as the original MV collector system, where the units in that system are assumed for simplicity to all operate at identical voltage.

The same procedure must be applied when the system is divisible into up to four component subsystems, each with its own independent connection to an MV collector bus. In this case, each subsystem furnishes its own aggregate with the equivalent collector impedance calculated as above.

The Applicant may propose an alternative aggregation method to the NSP and AEMO, who will jointly assess it. An alternative method is required in any case where the plant layout does not readily fit in either of the two categories above. As a matter of principle, it is noted that there is no one correct way to aggregate any collector system, and different principles such as equalizing losses (as per Muljadi et al), or reproducing an equivalent MV voltage diversity, will yield different results. All aggregation methods implicitly induce a degree of modelling error which must be acknowledged whenever comparisons are undertaken between modelled and true plant behavior.

General considerations?

For a generating system with many generating units, provision of the required aggregate model will be the primary method considered for wider power system studies and for AEMO’s own assessment tools.

Aggregate models should continue to provide access to the LV terminal bus quantities for each aggregate equivalent generating unit, including active power, reactive power and voltage magnitude. This includes EMT models that have been black-boxed. Figure 2 shows a high level example of how an EMT model that has been aggregated and black-boxed can provide access to terminal quantities.

For model validation purposes, the aggregated generating unit and aggregated generating system responses must conform to the accuracy requirements as specified by AEMO, where the individual aggregated generating unit model terminal quantities have a slightly moderated accuracy tolerance compared to the model connection point quantities. The procedure for R2 validation will involve collecting field measurements both for the aggregate generating system and for one representative generating unit for validation.

High voltage plant connecting directly to the transmission network is to be explicitly modelled.

Ref: AEMO guide

#AEMO #GenerationConnection #Windfarm #SolarPV #PowerSystemStudy