Traditional SCADA systems take basic status and analogue information and generate alarms, undertake logic sequences and allow remote circuit breaker operation (currently this is likely to be at primary substation level and above). At some point, if the goals of ANM are to be achieved, it seems likely that far greater levels of control, automation and data acquisition will be needed in the MV network below primary substation level.
The question of what level is or will be needed is largely beyond the scope of this report, however two different scales of active voltage controllers (as referred to in ER 12629) will be considered. The first is a micro-grid controller, aimed at niche pockets of the network containing DG and effectively acting as distributed intelligence controllers. The second is aimed at providing a broader level of functionality that can be gradually implemented over the whole network as an expansion of existing SCADA systems.
Currently, most AVCs controlling transformer OLTCs operate as stand alone control loops using static setpoints calculated in advance and then set and largely left alone, although most DNOs can set setpoints remotely from SCADA5, at least at the primary substation level. This system of voltage control (with or without the other various modifications considered in ETR 126 such as cancellation CTs), might not be sufficient when a DG is connected into the network under all normal load and generation scenarios as discussed above in relation to IVRs.
An active voltage controller takes a certain amount of real time measured information, a model of the network and some basic load profile data and then estimates load flows and voltages out on the network. With this information, it will dynamically adjust the setpoint of the transformer (or transformers) AVC or control other IEDs as appropriate, possibly including the DG, to keep the voltage within limits at all points in the area impacted by the controlled devices. The operation of a voltage controller is described more fully in ETR 126.
29 “Engineering Technical Report 126 - Guidelines for Actively Managing Voltage Levels Associated with the Connection of a Single Distributed Generation Plant”, Energy Networks Association, 2004
Two manufacturers in Great Britain are currently developing micro-grid controllers, Econnect with the GenAVC™ project and ABB with the AuRA-NMS™ project. The
GenAVC project, which is now commercially available, has been largely funded by the DTI and is a key part of the Martham Primary RPZ. The AuRA-NMS™ is being developed in conjunction with IFI funding and is still in development. A considerable body of literature is available regarding the GenAVC™ and ABB was approached for further information and responded to a questionnaire. Information from both sources is presented here.
Both are designed to be installed in the substations and control an area of the network (a
“micro-grid”). The AuRA-NMS™ will also perform dynamic system analysis, building on load flow analysis capabilities. Neither product has reached full commercial mode as yet, but both are currently undergoing extensive pilot trials and could be reasonably expected to be available within the 5 year timeframe considered in this report.
The ENMAC™ DMS/SCADA system is widely used amongst DNOs in the UK, and is a part of the Skegness RPZ. The manufacturers, GE Energy, have developed a Distribution Power Analysis (DPA) module to perform online power system analysis and state
estimation which interfaces between the existing SCADA dataset of real values and network data and an analysis engine. GE Energy responded to a questionnaire regarding this module, as well as more generally regarding the capabilities of ENMAC™ and ANM modules under development. This information has been also used for this report, and it is assumed that other control systems have or could have similar capability within the
timeframe considered for this report.
The DPA module has been extensively tested over several years, in particular with one DNO in Australia; however it is currently missing the final step of active voltage control, which is the functionality of actively managing the setpoints. GE Energy is developing that end to end functionality in an ANM module to interact with the existing systems. This system could also reasonably be expected to be available within 5 years.
In both cases, standard state estimation methods have been used, as they have been for many years at a transmission level, to estimate analogue values such as feeder voltages that are not currently measured in the field. The micro-grid controllers have been
developed specifically for the active voltage control function, whereas the DPA module is used by operators for a range of operational activities, including predicting the outcome of switching procedures.
6.3.2 Technical
Safety and Environment
Beyond ensuring that the individual components of the controllers meet appropriate safety standards, the key concern will be that the controller causes voltages at the customers’
terminals to exceed limits and cause equipment damage.
Unlike, for example, oil, gas, and chemical installations and thermal power stations, most control systems for electricity networks tend to be based on fairly low cost technologies and design principles and are rarely fully duplicated (excluding some protection systems).
Therefore, control logic must be designed to be failsafe to ensure that when any part of the control and communication systems fails, an unsafe situation will not occur. Both the micro-grid controllers and the central control modules have this capability.
Accuracy of the active voltage control system will largely depend on the accuracy of measuring and controlled devices, such as transducers and tap changers. This is therefore not different in principle from existing forms of voltage control at the primary substation and should be considered in the same way for each new installation.
Failure of the unit to operate as expected must also be considered. It is possible that the unit will drive the voltage in the wrong direction and exceed limits, for example because of an incorrect reading or unexpected network configuration. However, these risks already exist on the network, in particular during manual operations and it could be argued that the micro-grid controllers with their state estimation abilities reduce these risks overall.
Functionality
In general, studies30 and trials31 have shown that where voltage rise was a limiting factor for connecting DG to a feeder, active controllers can roughly double the amount of DG that can be connected to that part of the network without substantial reinforcement. Voltage problems on 11 kV networks are likely to be particularly relevant as at these levels transformer taps are often fixed. Therefore, the expected benefit from an active voltage controller for a particular area might be in the order of 2-4 MW per feeder of additional generation.
The range of the controller will be limited by the extent to which voltages on different feeders differ. If the controlled devices (usually the OLTCs at the primary substations) cannot physically be set to keep the voltage of feeders without DG within limits at the same time as the voltage on feeders with DG, then the active voltage controller solution will not be sufficient. Some other form of voltage regulator, such as an IVR, capacitor banks or an SVC or STATCOM might need to be implemented, either as a stand alone device or to be controlled in conjunction with the other devices by the controller.
Alternatively, on the occasions when the voltage rise was too great, DG might need to be constrained by the controller. Studies would be needed on a case by case basis to determine likely scenarios. As with the previous two ANM technologies discussed, these studies are likely to require a change in planning practices by the DNO, which should be easily implemented.
Centralised Control versus Distributed Control
The question of whether to employ devolved controllers at the substation level versus using centralised controllers depends on a number of factors. Those discussed below include:
Ease of maintenance following system changes
Standardisation of hardware and logic
Replication of hardware and logic
Speed of deployment
Need for and speed and reliability of communications
30 “Integration of Operation of Embedded Generation and Distribution Networks”, UMIST, Econnect, 2002
31 “Active Local Distribution Network Management for Embedded Generation”, Econnect, 2005
Network changes may need to be configured into the controller, presenting an ongoing maintenance challenge and also potentially requiring additional expertise for operations and maintenance personnel. This could be a particular problem at the 11 kV voltage level where changes to the network can be frequent and options for reconfiguring the network to restore supply following a fault are numerous. Modifications might be more manageable at the central SCADA level than at the distributed controller level, as changes to network configuration must be captured in network diagrams regardless of the presence of control.
One of the goals of the Aura-NMS™ project is to provide a repeatable, non-complex distributed solution for this issue32, however the project development is not yet far enough advanced to accurately assess whether this will be achieved.
Another consideration is amount of replication required for each new micro-grid. In both the centralised and distributed cases, the necessary measurement and control devices and the communications to the nearest data collection point (usually envisaged to be at the primary substation) will be needed. In addition, all of the standalone micro-grid
controller hardware must be replicated, whereas the centralised controller might only need, for example, an additional card on an existing RTU to feed data back to the control room.
However, where existing SCADA systems do not have the capabilities to easily incorporate active voltage control, localised, specialist controllers might be a more appropriate means of rapidly deploying active voltage control functionality. Similarly, if a DNO expects only a few micro-grids to be required based on the location and numbers of DG and does not see any significant benefits from the broader scale implementation of state estimation techniques, then localised controllers might be appropriate. In this case though, there is the risk of creating “legacy issues”, where a few numbers of unique devices, each highly customised and non-standard, are on the network needing specific knowledge transfer as personnel change.
Installation of a micro-grid controller at secondary or low voltage substations, where DNOs typically do not have any visibility of system voltages, could remove the need to install or upgrade communications to the SCADA following connection of DG. This assumes that the secondary substation transformers have OLTCs, without which the active controller has nothing to control, and this is frequently not the case.
It is likely that the normal status points and probably some information from the DG will still be required by the operator. Using a centralised controller will have some advantages in ensuring that all of the system data is readily available in a single place, so that asset maintenance, outage planning, system planning, fault level analysis, fault detection, trending and other functions can use a single, centralised data source. In addition, updating of the application software and hardware as well as the network configuration should be simpler.
Balanced against this is the faster control possible with a local, rather than remote
controller. Considering the various paths currently used to communicate between primary substation RTUs and the SCADA (described fully in reference 5), a complete control loop might take up to 15 seconds. This, coupled with the response time of the OLTC might be too slow to maintain voltages within limits, and would have to be considered on a case by case basis.
32 “Innovation Funding Incentive Annual Report”, Scottish Power, 2007
For example, this might not be a problem for most steady state conditions, as load generally changes slowly and, in the case of wind generation, maximum wind generation variation is unlikely to be more than 10%33 in the 15 second timeframe. Following a system disturbance, particularly a generation trip, there might be a need to rapidly change voltage settings to prevent prolonged temporary overvoltages on the network. However in general, there should be no need for a faster control loop because of the speed of the controlled devices. If there is a need for faster response times, upgrading the
communications link between the primary substations and the central SCADA system might be the solution where this could be economically justified.
One of the key considerations with regards to communication is reliability. If the controller is critical to maintaining voltages and the controller is located remotely, then the
communications to the controller will be critical. The system must already be failsafe in case of failure of the actual controller or other elements of the control loop, for example by causing any connected DG to go to a predetermined output in case of loss of
communications to the controller. However, the more unreliable communication links between the controller and controlled and measuring devices that are added, the more likely triggering the failsafe position becomes.
6.3.3 Operational
In each case considered, hardware and communication systems appear to use standard, mature technologies, and should integrate well with existing systems. The main design and maintenance concern, as stated above, will be the ease of updating the network models and other elements of the controller following changes to the configuration of the network. Standard logic and interfaces are needed to minimise customisation
requirements at each site. A detailed examination of the installation requirements for each device would be required before being able to assess this further.
Provided due care is given to the setup of the controller, its failure should not cause any negative impact on CML and CI measures (for example, on tap changer failure where DG setpoints reverted to “safe” values). Rather, the controllers should offer further
opportunities for improving CML in particular. CML was possibly a key driver for the original introduction and expansion of SCADA in some of the Great Britain networks, as it provided a means to reduce the time taken to identify fault locations. In addition, some DNOs have experienced considerable CML improvements using automated post-fault feeder restoration34, which could be added functionality of an active voltage controller.
6.3.4 Planning
Reliability and Consequences of Failure
The main consequence of a controller failing to operate as expected is likely to be a voltage excursion one or more parts of the network. This might be caused in a number of ways, such as:
Incorrect network data being input into the controller model
33 Based on the experience of SKM’s wind generation group
34 For example, EDF undertook a project to automate 1700 feeders with considerable improvements in CML – “Electricity Distribution Cost Review 2004/05”, Office of Gas and Electricity Markets, 2005
Insufficient feedback from measured values
Some novel feature of the network (such as a new FACTs type device or the DG) has not been correctly modelled in the controller
Some unexpected event or network configuration has occurred that the controller is unequipped to cater for
In essence, these causes of failure are not dissimilar to the types of problems that might occur with any system modelling of the network, for example to determine where
investment is required, to develop DG constraints for contractual arrangements or simply to determine the original setpoint for an AVC. One benefit of the active controller is that when the problem is found (preferably during commissioning tests), the solution is likely to be easily implemented. Another is that while the same level of conservatism could be built into the active voltage controller as that used now in system modelling, the controller has a feedback mechanism and therefore can be tuned to operate the network closer to voltage limits with greater confidence.
Commissioning of such a system, like any control function, will have to be carefully considered and comprehensively carried out. Creation of all likely network conditions for testing might be difficult, as with any network equipment. For this reason, the GenAVC is undergoing extensive trials in two sites, connected at first in open loop mode with control outputs carefully monitored. Initial testing of the DPA module of ENMAC™ at an
Australian site was similarly carried out over a 12 month period in open loop mode.
Such extensive testing is not likely to be commercially feasible if these controllers are to become standard practice, and therefore the extent of the difference in control logic between the project requirements and the extensively tested version of software must be considered on a case by case basis.
GE Energy report an initial accuracy of 90% of the DPA module estimated network values when compared against directly measured values. This accuracy was achieved despite the state estimation algorithm being supplied only with single load values collected
quarterly from secondary substations during inspections. After a period of monitoring and loop tuning by control room operators, the accuracy was considerably improved. Econnect report that GenAVC has operated for considerable periods in open loop mode showing that correct control actions would have been taken in closed loop mode. Therefore, in both cases it can be assumed that the state estimation functionality will provide operators with considerably more information than they current have access to.
Cost
The main benefit of installing an active voltage controller in a section of the network with connected DG will be to avoid large capital costs (usually attributed to the DG developer as part of the connection costs) of, for example new circuits at higher voltages. When comparing the capital costs in such a case, the controller option is likely to be substantially cheaper. However, the first such arrangement each DNO installs is likely to involve extra internal costs in adjusting to a new way of managing voltage control, and will require personnel to step out of the normal roles in order to implement these changes. A broader change to the main SCADA system will require additional training and participation by control room staff.
6.3.5 Other
To qualify for IFI funding for an active voltage controller based on the GenAVC™ or AuRA-NMS™ system might be difficult unless the DNO could prove that the “application domain or operating context” is “new or unexplored”. This is possibly the reason that only one of the GenAVC™ trial sites was registered as an RPZ. However, the rules for the IFI actively encourage collaboration between DNOs, and so it is theoretically possible that a DNO not currently involved in AuRA-NMS™ might join the project.
Another means of qualifying for either IFI or RPZ using GenAVC™ might be to incorporate the controller with a piece of primary plant with which it had not previously been trialled, such as an IVR or a FACTs device.
As active voltage controllers for distribution networks are not a fully established
technology, and each new manufacturer is likely to propose a somewhat different control algorithm or methodology at this stage, any other controller should qualify. This should include extension of the capability of a DNO’s existing SCADA system to provide such
technology, and each new manufacturer is likely to propose a somewhat different control algorithm or methodology at this stage, any other controller should qualify. This should include extension of the capability of a DNO’s existing SCADA system to provide such