The influence of micro-CHP units on a distribution network1 December 2007
Nuon has been enlisted by a consortium of Dutch energy companies to assess the effect on what would normally be a passive distribution network of the introduction of numbers of CHP micro-generation units. This assessment has come to be known as the Paddock Trials.
Distributed combined heat and power generation is seen as a key weapon in the drive for greater energy efficiency in the generation of electricity. However it also raises a number of important issues. Prominent among these is the effect of small generating units on the operation of a distribution system.
In the standard hierarchical electricity supply network, power is generated centrally and transported to users via a transmission and then a distribution network. Power management and system balancing is carried out at the transmission system level and the distribution network is conceived as a passive network. Introducing distributed generation which feeds power into the system at the distribution network level can upset the balance within the distribution system, affecting its stability and key operational parameters.
The smallest distributed CHP units likely to be encountered are domestic systems designed to provide both power and hot water to a single household. These are not commonplace today but a number of companies are designing units that will be used for this purpose in the future. Among these are systems which use Stirling engines to generate electricity.
To explore the interaction of the micro CHP systems with the low voltage grid the Dutch distribution system operators Continuon, Eneco NetBeheer and Essent Netwerk together with GasTerra, a gas trader, and the Smart Power Foundation (a collaboration of boiler companies) commissioned a set of trials of such units. Their goal was to examine the impact of micro CHP on the low voltage grid in order to prevent problems with the introduction of these boilers into the market. These sponsors enlisted Nuon Tecno (part of the Dutch electricity company Nuon) to carry out these trials.
In 2006 Nuon was given the opportunity to test a first group of units of this type. This first trial involved 48 WhisperGen micro CHP units provided by the New Zealand based company Whisper Tech. These formed the basis for what became known as the Paddock Trial 1.
In Paddock Trial 1 the 48 units were installed in a test rig designed to simulate a small section of a low voltage distribution network supplying power to group of households. Subsequently Nuon also had the opportunity to test 18 units built by Microgen. These were tested alongside 24 of the original WhisperGen units in Trial 2.
Stirling engine CHP units
The Stirling engine incorporated in these micro CHP units is probably the oldest type of heat engine in use today. It was designed by a Scottish Presbyterian minister in 1816 and the engine is unusual among modern heat engines in that it operates in an entirely closed cycle with the heat to drive the cycle applied externally. Stirling engines are capable of high efficiency and extremely quiet operation and require little maintenance. However they tend to be expensive to construct.
The WhisperGen Mark IV units used in these trials employ a Stirling engine with four closely coupled cylinders arranged axially, leading to low vibration and smooth torque. Heat to drive the engine is provided by burning natural gas, with waste heat captured to generate hot water. When the engine is used to drive a generator and a typical unit can provide an output of 1kWe and a heat output of 7.5kW-12.0kW.
The Microgen units also employ a Stirling engine as the prime mover, and the engine in the trial provides around 1kW of electricity and 5 kW of heat output, but can incorporate supplementary burners for additional heating capacity. The main difference between the two is that the Microgen unit is equipped with a linear synchronous generator rather than the more conventional rotary asynchronous type fitted to the WhisperGen.
The Microgen system is based on technology licensed from US company Sunpower and was being developed by UK company BG (formerly British Gas). However BG abandoned the project earlier this year. Since then the intellectual property rights to the system have been bought by a Netherlands consortium, the members of which are interested in taking the technology forward individually as the basis for domestic CHP systems for the European market.
The Paddock simulation
The Paddock trials were designed to examine the behaviour of the micro CHP units in a simulation of a low voltage network supply system. The units were all housed in a single building in a electrical system shown in Figure 1. The system comprised a low voltage transformer feeding a low voltage cabinet supplying two separate low voltage distribution feeders. The low voltage feeders were fitted with disconnectors so that they could be connected into a ring or as a single feeder.
Each of the two feeders ran for 65m before being connected to 24 micro CHP units separated by distances which were designed to simulate a housing estate. The branches for each unit were separated by 7m and each branch from the feeder to the micro-CHP unit was 10m long. Figure 2 shows two possible configurations for the low voltage network used in the trial. The 48 units were arranged in four rows of 12, with two rows representing one feeder, as shown in Figure 3. A view of the units photographed during the tests is shown in Figure 5.
The units were all controlled from a central computer which allowed each unit to be switched on or off individually. The software also allowed scripts to be written to automatically switch units on or off in a systematic manner to simulate particular extraordinary or fault situations. No attempt was made, however, to simulate real households since the aim of the study was to find out how the units would behave under extreme conditions and such conditions would occur infrequently in a real-life simulation.
During both Paddock trials the system was fitted with four power quality measurement points (Figure 4) where transient analysers were used to record voltages and amplitudes on the low voltage feeder. One analyser was situated at the junction between the transformer and the feeder cabinet while the other three were between branches on the feeders.
One of the points of the trial was to see how the units behaved during a fault such as removal of the low voltage supply or a short circuit. In was initially imagined that whatever the behaviour of a single unit under these conditions, this would be amplified 48 times when 48 systems were operating together. In practice the behaviour was found to be more complex than this, with some cancelling occurring so that the overall effect was smaller than expected. In order to explore this aspect more thoroughly, transient quality measurements were taken on the feeder between every pair of units during trial 2 so that the interaction of every pair could be recorded.
Paddock trial 1 involved 48 WhisperGen units. However there were only 18 Microgen units available for trial two. For this trial all 24 WhisperGen units were removed from one feeder and the Microgen units put in place of 18 of them. In order to balance the two feeders, only 18 of the 24 units on the second feeder were actually operated on in trial 2.
In order to calibrate the system and establish some baseline operational parameters, a series of simple tests were carried out, measuring the start-stop behaviour of the units under hot and cold conditions. Unlike some types of generating unit, Stirling engines have to warm up before they can operate. This can take several minutes. For this reason WhisperGen CHP units cannot be stopped and started under thermostatic control like a gas-fired domestic boiler. Once started, they should run continuously for optimum performance.
Because of the relatively quick response of the Microgen engine, on-off control is possible. With the higher electrical efficiency yielding a lower heat output, on-off cycling times can in practice be long. In field tests typical cycling times of one hour have been measured with minimum variation in room temperatures and such low-switching behaviour leads to better efficiencies.
Both types of micro-CHP unit use gas to heat the engines during startup. Both are also fitted with a dump or a stall resistor which is there to take up surplus energy when, for example, the unit is unexpectedly disconnected from the grid. The WhisperGen units test this resistor during startup and as a result initially draw current from the grid. The startup behaviour of both is compared in Figure 6, which shows the WhisperGen startups in the top set of recordings and the Microgen startups in the bottom set. The Microgen engines stall within 3 cycles of an unexpected disconnection of the grid, so dissipated energy is negligible.
The Microgen hot start behaviour is virtually identical to its cold start behaviour. However the WhisperGen tends to start more quickly when hot and reach its operating output more quickly than from cold.
Another set of tests carried out on the two sets of units involved putting a resistive load on the two feeders so that the power drawn would be equivalent to one third, two thirds and full power on each unit. The two sets of units in trial 2 showed small step changes in voltage output as power drain was increased. The Microgen units showed much greater variance of output from unit to unit than the WhisperGen units, but this is considered to be a matter of factory calibration.
One of the most important tests was to examine the engines’ behaviour under islanding conditions when the main grid connection was interrupted. Micro CHP units of this type will generally only be able to supply a part of a household's demand, most of the time, with the rest being drawn from the grid. They must, therefore, be grid connected. The units may be able to feed power to the grid under certain circumstances.
In order to test islanding behaviour, 23 of the units were set up to supply around 700-800W each into a resistive load. Then the main feeder was disconnected from the low voltage transformer, leaving the units operating without any distribution system connection.
When the circuit was opened, both types of CHP unit continued to supply power. Voltages from both increased from 225V to around 235V when the circuit was opened and power output rose as much as 10% on some Microgen units.
The most intriguing behaviour, however, was found in the AC frequency recordings. Normally the grid frequency was superimposed on both units. However when the circuit was opened the synchronous linear generators of the Microgen units took over frequency control so that the supply frequency became that of their generators, 49.9Hz. Since the WhisperGens are rotary machines with no resonant frequency it is not possible for them to operate alone in island mode.
The second major fault test involved examining the behaviour of the units when all three phases were suddenly shorted out to earth for 70mS, simulating the conditions that would normally cause a fuse to blow. In this case the initial response of the two types of unit was quite different. Other tests examined the behaviour with short circuit durations ranging from 12mS to 200mS.
The WhisperGen units continued to maintain their output voltage during the short circuit, though with some large current swings contributing to a significant fault current. When the load was reconnected they resumed functioning normally. In the case of the Microgen units, however, the initial result of the short circuit was to cause their output voltage to fall away completely after a few cycles so that their current output died. Figure 7, which compares the two, shows this behaviour clearly.
This unexpected response appears to be due to the way in which the control circuits in the Microgen unit are designed. After some changes to the units’ settings they were then able to operate continuously, supporting grid stability without adding to the fault current and the problem was eliminated.
The Paddock trials 1 and 2 represent an important test of the performance of micro CHP units on a distribution network. The trials looked at the grid side rather than the household side behaviour of the units and the results have important implications.
Perhaps one of the most significant findings relates to behaviour when the units become islanded, intentionally or unintentionally, when the power output of the units is in balance with the power demand. As was seen above, it is possible that they will maintain their output which means that the islanded distribution system remains live as a balanced system. It is normal practice when working on distribution systems to assume that the low voltage network will not be live if it has become disconnected so this finding could have safety implications. Clearly, technicians need to be aware that the distribution network may remain live, even when disconnected. But this is the result of a test in the controlled environment of a laboratory. It remains to be seen what this means in reality because the situation only arises if output and demand are balanced at the moment of disconnection.
From a network operational perspective the short circuit behaviour, and particularly the dynamic peak contribution, is also important. In this case the Microgen units shut down after a very few cycles and contributed very little to the fault current. The WhisperGen units remained operating, however, and each unit provided a significant fault current, (3 to 8 times their nominal current). However the contribution from each unit was slightly out of phase with the others so that the cumulative fault current was actually less than might otherwise have been anticipated.
Another finding was that the units are capable of increasing their voltage and power output very rapidly, more rapidly than traditional transformer tap changers would be able to respond were this superimposed on the distribution network. This could present problems if a large group of such units were all restarted at the same time. Again the question is what this means in reality. Under normal operating conditions it is very unlikely that all units would start at the same time.
The two sets of tests that comprised Paddock trials 1 and 2 provided an important benchmark of the behaviour of this type of unit for future reference. In the light of this the performance of other units can be assessed, the results being used to tailor their performance so that they behave as benignly as possible from a distribution grid perspective.