The developed planning procedure

A DG interconnection planning procedure regarding voltage issues (voltage level and fast voltage transients) is determined in [P7] and [P8]. In addition to statistical planning, also worst case studies are needed to make sure that technical constraints of the network are never overstepped. Network voltages need to remain at an acceptable level throughout the year and the fast voltage transient at generator connection or disconnection cannot exceed the set limits. The developed planning procedure is depicted in Figure 5.2.

At the first step of the planning procedure, input data is obtained. Network data and hourly load curves can be obtained from the NIS and generator data including for instance the reactive power operation and start-up current of the generator are obtained from the potential energy producer. Statistical production curves are created using the method of [121]. Also minimum and maximum production curves are needed for the worst case calculations. The minimum production curve is, at present, always a zero production throughout the year regardless of the type of the DG unit because it cannot be guaranteed that the DG unit produces at a specified hour. If, however, a large number of different types of DG units is

connected to a network, the probability of zero production can become so small that the minimum production curve needs to be altered to larger values than zero production.

Figure 5.2. The planning procedure concerning voltage issues when a new DG unit is connected to an existing distribution network.

The maximum production curve varies depending on the type of the DG unit and on whether a firm or a non-firm connection is used. If production curtailment is allowed, the maximum production curve can also be set to zero throughout the year because if voltage rise becomes excessive at some hour, production curtailment can be used to restore network voltages to an acceptable level. If a firm connection is used, the maximum production curve is usually a flat curve representing nominal power production throughout the year. In case of, for instance, a CHP plant, the plant might not be used at summer time and, hence, the maximum production curve can be zero during summer months and at the nominal power at other times.

At the second step of the planning procedure, worst case calculations are conducted. These are needed to verify that the network operates acceptably throughout the year. The worst case

Obtain network data, hourly load

curves and generator data Create production curves (statistical, minimum and maximum) Step 1: Input data

Step 2: Worst case calculations Maximum load/

minimum generation

Plan network reinforcements or discard the control approach in question Voltage quality acceptable?

Step 3: Statistical planning

Calculate load flow Set excess probability

of load to 50 %

Select statistical production curve

h < h_end h = h + 1

All voltage control methods examined?

Compare different control approaches and select the most cost-effective one

no

Changes to network data

no

yes

yes

no yes

Network reinforcement?

no yes Select the studied voltage

control strategy

Fast voltage transients at generator connection

and disconnection Maximum load/

maximum generation

Minimum load/

maximum generation

Select new voltage control strategy Use original network data

calculations include calculating the fast voltage transients at generator connection and disconnection and calculating the voltage levels at three extreme loading conditions.

The transient voltage variation at generator connection and disconnection can be determined by calculating load flow with and without the generator. In these simulations, the voltage regulating devices such as main transformer tap changers should be disabled and the loading situation should be the one in which the DG unit causes the largest voltage variation in the network. This is usually the minimum loading condition. To be able to perform these studies, the start-up current of the generator needs to be known.

The voltage level of the network also needs to be studied during the worst case calculations.

Three loading conditions are examined at this step: maximum load/minimum generation, maximum load/maximum generation and minimum load/maximum generation. These studies are used for network dimensioning purposes and, therefore, load powers that are not gone below (minimum loading) or exceeded (maximum loading) with some probability, are used.

In voltage drop calculations, excess probabilities of around 10 % are often used [7]. This might be a suitable probability for these studies as well. Hence, the excess probability of loads can be set to 10 % when the maximum loading is determined and to 90 % when the minimum loading is determined.

The maximum load/minimum generation case is the case which is, at present, studied when distribution networks with no DG are planned. The dimensioning factor is voltage drop and, therefore, the substation voltage is set to its lowest possible value in these calculations. The minimum load/maximum generation case is the case which is, currently, used when interconnection planning of DG units is conducted (see chapter 5.1). In this case, voltage rise is the limiting factor and, therefore, the substation voltage is set to its highest possible value in these calculations. The maximum load/maximum generation case is needed because the difference between network maximum and minimum voltages can be largest in this case.

Hence, if coordinated substation voltage control is used, it is possible that the control algorithm is able to keep all network voltages at an acceptable level in the minimum load/maximum generation case but at the maximum load/maximum generation case the network voltages cannot be normalized by using only the substation voltage control because of the large difference between network maximum and minimum voltages.

After the worst case calculations, it is checked whether the transient voltage variations at generator connection and disconnection are acceptable and whether the network voltage level remains acceptable during all hours of the year when the selected voltage control method is used. If the voltage quality remains acceptable, statistical distribution network planning is conducted. If any of the network’s technical constraints is overstepped, network reinforcements are planned or a new voltage control strategy is selected and the worst case calculations are redone.

Worst case calculations and statistical calculations are conducted for different voltage control strategies and the outputs of these calculations are used to select the most suitable method for a particular case. The usage of the developed planning procedure is demonstrated in [P8].

The statistical planning procedure has been used in this thesis to study issues regarding network voltages, but a similar planning method could also be used when for instance thermal constraints are examined.

5.2.2.1 Development needs of network information systems

Some development of the current NISs is needed to be able to execute the statistical planning procedure introduced above. The Nordic NIS already includes network data (feeder impedances etc.) and loads are modelled using hourly load curves [89] and customer data obtained from CIS, but models for production curves and active voltage control methods are not available. These models need to be added to the NIS to enable the conducting of the hourly load flow calculations. Steady state calculation methods are used in the NIS and, hence, also models of DG units and active voltage control methods need to be simple enough.

At present, DG is modelled in NIS load flow calculations as a negative load with fixed real and reactive powers. This is adequate when the DG output is independent of the network state but when the DG unit is operated for instance in voltage control mode, its reactive power output depends on the terminal voltage and, hence, fixed reactive power output cannot be used. Therefore, the DG models need to be extended to enable modelling of different real and reactive power control strategies. Time domain modelling is not needed but a simple steady state model is adequate. For instance, if the DG unit operates in voltage control mode, only the droop curve describing the dependence of reactive power output on the terminal voltage needs to be added and the actual implementation of the control does not need to be considered.

Other active voltage control methods also need to be modelled using a simple enough description for NIS steady state calculations. Methods based only on local measurements can be quite easily modelled: only the dependencies between the measured variable and the controlled variable need to be described. Modelling of coordinated methods can require more work depending on the complexity of the control method. If active voltage control is implemented as a part of the DMS, and the NIS and the DMS are highly integrated, the DMS models can be directly utilized in NIS calculations.

The planning procedure of Figure 5.2 could be quite easily automated in the NIS. The different steps of the planning procedure could be implemented as separate functions. Load flow calculations for every hour of the year are needed in steps 2 (worst case calculations) and 3 (statistical planning). In step 2, hourly load flow calculations are needed when the calculations are conducted for the first time because the hours of maximum and minimum load are not known. In the following simulation rounds when different voltage control strategies are studied, the results of the first round can be utilized and simulations can be conducted only at the previously determined maximum and minimum load hours. In step 3, hourly load flow calculations are conducted at every simulation round. Conducting the load flow calculations by hand for every hour is not practically possible because there are 8760 hours in one year. Hence, the processes of steps 2 and 3 should be implemented as their own functions that would automatically conduct the required simulations for the whole year.

Implementing these functions would be quite easy because the already existing NIS

calculation functions could be utilized and only the loops implementing the calculations throughout the year need to be added.

A planning procedure for protection planning when DG is interconnected to a distribution network is presented in [10] and its implementation as a part of the NIS is discussed. If also the planning procedure regarding voltage issues proposed in this thesis was implemented as a part of the NIS, the DNOs would be able to conduct almost all DG interconnection studies using the NIS.