Published by M.Sc Marko Pikkarainen, M.Sc Pekka Nevalainen, Dr. Pertti Pakonen, Prof. Pekka Verho, Tampere University of Technology, email@example.com
Two matters have strongly affected use of energy in Finland during recent years. The one is the greenhouse effect and the other is the energy price which has risen. These points have brought to the market many new devices that are more energy efficient, for example, heat pumps and energy saving lamps. Also the reduced manufacturing costs of devices have brought new kinds of loads available in the market. These new devices have changed the usage of electricity and usage of the distribution network. The distribution network has been designed to fulfil different kind of electricity usage need. That is why more power quality problems may occur in the distribution network in the future.
This paper will describe some power quality problems caused by the usage of ground source heat pumps and a wood splitter. This examination is based on the measurements that have been carried out in the real distribution network at low voltage level in Finland.
The examination showed that power quality problems may appear when using these typical new loads. Especially one big problem is a starting current of a wood splitter. Because of high starting current the voltage will drop and it may cause flicker.
During recent years, the use of energy has changed because new kinds of loads have been connected to the distribution network. Three main issues have driven this change of loads. One is the green house effect and second is the energy price, which has risen, and third is the reduced manufacturing costs of devices.
The main act in preventing the green house effect and global warming is to decrease carbon dioxide emissions. Because of this more and more loads, which are reducing carbon dioxide emissions, have entered the market. For example compact fluorescent lamps are one group of such loads. Replacing incandescent bulbs with compact fluorescent lamps is one such act that should reduce carbon dioxide emissions because of the efficient light produce of compact fluorescent lamps compared with incandescent bulbs. The result of European Commission Regulation number 244/2009 is that incandescent bulbs will be gradually phased out from the market. [1, 6]
The increasing energy price has affected the use of energy so that customers have invested in devices which reduce costs and energy consumption. One good example of this kind of behavior is to replace an oil heating system with a heat pump or to add an air-to-air heat pump in complement electric heating. In Finland support from government has speeded up this change . Figure 1 shows the number of installations of different types of heat pumps in Finland during the years 1996-2008. As shown in Figure 1 the number of heat pumps has grown very rapidly during past few years . The growing trend has been the same all over Europe. The overall percentage of heat pumps of all heating types is not very massive in Europe but for example in Sweden heat pumps are the most common heating system in single-family houses with an approximately 34 % share of all. 
Different heat pumps may have different effects on electrical energy consumption and to the way electricity is used. For example, if an oil heating system is replaced with a ground source heat pump, the overall consumption of electrical energy of a house will increase, but if an air-to-air heat pump is added to complement electric heating the consumption will decrease. However in both cases if the earlier consumption of the primary energy source is compared with the new consumption of electrical energy the consumption is decreased because most of the heating energy of heat pump is coming from ground or from air. Because of decreased overall primary energy consumption also carbon dioxide emissions are decreased depending on how electricity is produced. The greater usage of fossil fuels in electricity production the greater cutting can be achieved using heat pumps.
Heat pumps are a good example of loads which have become more common because of technical development and reduced manufacturing costs. Nowadays in Finland an air-to-air heat pump costs about 1200-3500 € including installation which is quite feasible price in Finland .
Ground source heat pumps cost more because it will need a ground circuit and the device is bigger in the power scale. Another example of a load which is becoming more and more common because of a cheap price is a wood splitter. Wood splitters are used for splitting thick woods to smaller ones so that woods can be used in fireplaces or in sauna stoves. Especially the cheaper versions of wood splitters that are designed for regular customers are very tempting devices because of the easiness of wood splitting.
These new loads are changing the use of electrical energy. Even though these loads may have a favorable effect on overall energy consumption some power quality problems may occur when using these loads. This is mainly caused by the new electrical characteristics of these loads. The planning principles of distribution networks are becoming old-fashioned and do not always fulfill the requirements of these new loads. From the power quality point-of-view the trickiest part will be the commonness of these loads because it means that power quality problems are also becoming more common.
This paper will study some power quality problems caused by the use of ground heat pumps and wood splitters. Devices have been selected to this study based on power quality complaints received by one distribution utility. The study is based on practical case study measurements which were carried out in real distribution networks in Finland. In the paper there is first a theoretical examination about power quality problems and previous mentioned loads. This is followed by a description of case study measurements carried out. Finally there are results and conclusions of those measurements.
II. THEORETICAL BACKROUND OF POWER QUALITY
Power quality is defined as “Set of parameters defining the properties of power quality as delivered to the user in normal operating conditions in terms of continuity of supply and characteristics of voltage (symmetry, frequency, magnitude, waveform)” . In this paper, we are observing power quality in terms of quality of voltage. The limits for voltage quality are defined in standard EN 50160 Voltage characteristics of electricity supplied by public distribution networks. The standards object is to define and describe characteristics of the supply voltage concerning: frequency, magnitude, wave form and symmetry of the line voltages. “These characteristics are subject to variations during normal operation of the system because of changes of load, disturbances generated by certain equipment and the occurrence of faults which are mainly caused by external events”. Variation of the characteristics is random in time and location. Therefore on small number of occasions the limits can be expected to be exceeded. 
The standard EN 50160 defines the characteristics of voltage in low voltage and medium voltage networks . Because we are interested power quality problems caused by devices which are connected to low voltage networks only the characteristics of voltage in low voltage networks are considered in this paper. The examination is focusing on a flicker caused by rapid voltage changes, voltage levels and voltage dips because those are a very common cause for power quality complaints as shown in Figure 2. Figure 2 presents the distribution of power quality complaints in one distribution utility in Finland during the years 2003-2005. About 70 % of all power quality complaints were caused by voltage changes.
The other reason to focus only on flicker and voltage levels, when studying power quality, is that devices very commonly cause these disturbances. This mainly happens because a device requires power to operate. In addition voltage, a device will need current from the grid. This current will cause a voltage change over the impedance of a distribution network. This is seen in Equation 1. Depending on the connection of a device it may have different effects to phase voltages. If the connection is a three-phase connection and connection is symmetrical, all phase voltages will experience the same kind of effect voltage drop or voltage rise. If the connection is a single-phase connection, every phase will experience different kind of voltage change because of the star point displacement for example one could rise and the others drop.
ΔU = vector of voltage change
ZN = vector impedance of distribution network
IDEV = vector device current
The standard EN 50160 defines permitted levels for a flicker so that 95 % of long term flicker severity in any week should be lower or equal than 1. For voltage levels, standard defines that 95 % of the 10 min mean r.m.s. values of the supply voltage shall be within the range of Un ± 10 % during each one period and all of 10 min mean r.m.s. values shall be within range -15 % < Un < +10%. Exceptions for voltage levels can appear for example in remote areas with long feeder lines or not connected to a large interconnected network. In these cases voltage levels could be outside previous mentioned range but a customer or user should be informed of the conditions. 
III. THEORETICAL BACKROUND OF HEAT PUMP AND WOOD SPLITTER
In both devices, heat pumps and wood splitters, the source of power comes from induction motor. This component is the most significant component in these devices from the current usage and power quality point-of-view. In wood splitters the motor is usually a single phase induction motor. In heat pumps the motor can be either a single phase or a three phase induction motor. The connection type varies with different type of heat pumps. In bigger heat pumps, ground source heat pumps and air-to-water heat pumps, the induction motor is three-phase connected. In others heat pumps the induction motor is typically single-phase connected, because those heat pumps are smaller.
The biggest impact of an induction motor on the distribution network appears when the motor is started. The induction motor takes a high starting current. High starting current causes voltage change over the impedance of the distribution network. This phenomenon is seen in Figure 3 and equation 2. Figure 3 a presents an equivalent circuit of a polyphase induction motor and figure 3 b presents an approximate equivalent circuit of a polyphase induction motor. The approximate equivalent circuit is based on assumptions that the reactive component of impedances z1 and zm is much greater than the resistive component and the voltage E2 is only little smaller and nearly in the same phase with voltage V. These assumptions are valid in conventional induction motors in the normal running range. Equation 2 can be defined from figure 3 b. From Equation 2 can be seen that when the slip of the induction motor is small the motor takes a high current from a network. The slip is 1 at the moment of starting the induction motor and will decrease close to 0 after a starting. Single phase induction motors have a different equivalent circuit and equation for current taken from network compared to polyphase induction motors. Nevertheless the high starting current effect is similar to that in polyphase motors. 
r1 = resistance of stator
x1 = leakage reactance of stator
xm = magnetizing reactance
r2 = resistance of rotor
x2 = leakage reactance of rotor
s = slip of motor
The starting current causes remarkable power quality problems when using induction motors because current reaches highest value at the beginning of start up and it won’t fluctuate much during normal operation. From power quality point-of-view, the critical factor is how often the motor is needed to start up. If the start up frequency is very high more voltage changes will appear. For the heat pumps length of the running cycle depends on the need of the heating energy, the dimensioning and parameter settings of a heat pump. If the heating power demand is close to the nominal heating power of the heat pump, the pump may run long times continuously. If the heating power need is clearly lower than the nominal heating power of the heat pump, stopping and reclosing of the pump will appear. The time between the stopping and reclosing the pump depends on the restrictions of the process and, for example, for one ground source heat pump the shortest time between stopping and restarting the process is 10 minutes according to a heat pump supplier.
For wood splitters starting up frequency varies depending on the operation logics of the device. Basically there are two operation logics to move hydraulic piston of the wood splitter. One is to perform all piston movements with hydraulic control when the induction motor is running. The other is to do the pressing with a hydraulic control when the motor is running and the backward movement with a spring when the motor is stopped. Second logic means that the repetition frequency of starting ups will be very high. In our measurements we had a wood splitter that needed to start up again every time a new wood was split.
IV. DESCRIPTION OF PRACTICAL CASE MEASUREMENT STUDY
In our study, two practical case measurements were made in Finland one in Lempäälä and the other in Tampere. In both cases the scope and the environment were bit different. Dranetz PX-5 power quality analyzer and Dranetz 4400 power quality analyzer were used as measuring devices in our study. In this chapter both practical case measurements will be described in depth.
Lempäälä rural area network
In Lempäälä the scope of the study was to explore power quality problems caused by a wood splitter in a rural area network. The wood splitter was single phase device and the nominal power of induction motor of the wood splitter was 2.2 kW. Operation logic of this wood splitter was that it needed to start up again every time a new wood was split. Feeders in this low voltage network were mainly aerial bundled cables called AMKA with cross-sections from 70 mm2 to 35 mm2 and the rated power of the 20/0.4 kV transformer feding the low voltage network was 200 kVA. In this low voltage network computational single phase short circuit currents at customers varied from 1400 A to 148 A. We decided to study power quality problems caused by a wood splitter in three locations in which measured single phase short circuit currents were 146 A, 275 A and 350 A at customer supply terminal. There were 10 m extension cord between a customer supply terminal and a plug point of a wood splitter so single phase short circuit currents were 136 A, 200 A and 233 A at a plug point of device. Figure 4 A shows an overall picture of low voltage network in Lempäälä. Measuring locations were situated along feeders 1 and 2. When a wood splitter was operated measurements were made in three places: one in a plug point of device, one in a customer supply terminal and one near the transformer. In addition when the wood splitter was operated in the location where the short circuit current was 146 A measurements were performed also in the location with a short circuit current of 275 A because those were located along the same feeder. Measured quantities were voltage and current waveforms and quantities defined in standard EN 50160 with an exception that a measuring period of short term flicker severity was 5 min.
Tampere urban area network
In Tampere the scope of the study was to explore power quality problems caused by ground source heat pumps in an urban area network. Feeders in this low voltage network were mainly underground cables with cross-sections from 300 mm2 to 120 mm2 and the rated power of 20/0.4 kV transformer was 315 kVA. In this low voltage network computational single phase short circuit currents varied from 9,7 kA to 445 A and three phase short circuit currents varied from 10.8 kA to 1110 A so the network was quite strong. At the one end of this low voltage network two terrace houses changed their shared oil heating system into a separate ground source heat pump systems. The installation was made so that two ground source heat pumps, nominal heating powers 25 kW and 36 kW and maximum electrical powers 9,9 kW and 13,2 kW, were installed to both terrace houses. Starting of all heat pumps was direct on line starting so every time heat pumps were started a high starting current appeared. This place was selected to this study because some customers in both terrace houses complained about flicker. Overall picture of low voltage network, customer supply terminal short circuit currents and measuring points are illustrated in figure 4 B. Measured quantities were voltage and current waveforms and quantities defined in standard EN 50160.
V. PRACTICAL CASE STUDY RESULTS
This chapter presents the results of practical case studies. This chapter is headlined similarly as the previous chapter so that it is easy to follow results.
Lempäälä rural area network
Power quality problems caused by a wood splitter were remarkable at the customer end. Every wood splitting produced high starting current compared with the short circuit current and a remarkable voltage dip in the phase in which the wood splitter was connected at the customer supply terminals. This is why the biggest problems appeared in flicker and in number of voltage dips. The waveform and the RMS value of a starting current of one start up of a wood splitter are illustrated in Figure 5.
It was detected that phase voltages of other phases than the phase in which the wood splitter was connected were raised at a customer supply terminal. This is due to a star point displacement of low voltage network when using single phase devices as predicted in Chapter 3. This effect is illustrated in Figure 6. Figure 6 shows phase voltages at customer supply terminal during start up of a wood splitter. Because of this effect power quality problems also appeared in other phases than the phase where a wood splitter was connected. Also Figure 6 shows that the phase voltage in connection phase drops so dramatic that every start up produced a voltage dip according to standard EN 50160 .
Overall network impact results of practical case study in Lempäälä at customer supply terminal are summarized in Table 2. In measurements 1 and 3 the wood splitter was connected in phase L2 and in measurement 2 the wood splitter was connected in phase 1. Table 2 shows very dramatic short term flicker severity index increase at customer supply terminal in every phase when using the wood splitter at a customer installation. This means very annoying flicker and it also means that even short use of a wood splitter will exceed the limit Plt=1 in long term flicker severity calculated with definition in standard EN 50160 . It should be noticed that an electrical chainsaw was used at same time when the wood splitter was operated. This increased little a short term flicker severity index. The effect of an electrical chainsaw to phase voltage was clearly smaller than the effect of a wood splitter.
Table 2. Overall results of a practical case study in Lempäälä
In addition of Table 2 results it was noticed that when the wood splitter was operated in measurement place 1 also power quality problems were recorded in measurement place 2. Geographical distance between these two places was 250 m. Operation of the wood splitter raised the short term flicker index of the other measurement place up to 7,2 in phase where a wood splitter was connected and up to 2,8 and 1,1 in other phases even though a wood splitter was operated in measurement place 1. Even though remarkable power quality problems appeared at customer end no power quality problems appeared at transformer. One way to prevent power quality problems mentioned in this chapter is to use only wood splitters of which piston movements are controlled with hydraulic control while the induction motor is running continuously.
Tampere urban area network
In Tampere four heat pumps were operated in same low voltage network. The measurement period was one week. During this period the mean temperature of a day was 3…9 °C and at night the temperature fell under 0 °C. This meant that heating power need was not near heating capacity of pumps so start ups and stops of pumps should appear. Installations of pumps were made so that a bigger pump heated only a water circulation of radiators. A smaller pump heated mainly use water but could support a bigger pump to heat a water circulation of radiators. Because of a direct on line start up of pumps high starting currents were detected. Starting currents of bigger pumps were 220…230 Arms and duration approximately 4 cycles. Starting currents of smaller pumps were 185…195 Arms and duration approximately 3 cycles.
Running cycle of pumps varied between terrace houses. Bigger pump of customer 1 ran typically from 1 h to 3 h 20 min. Time between two start ups varied from 1 h 40 min to 4 h 10 min and the average time between two start ups was 2 h 20 min. The total number of start ups was 73. Bigger pump of customer 2 ran typically from 50 min to 11 h. Start up times of bigger pump of customer 2 varied from 2 h 20 min to 12 h 10 min and the average time between two start ups was 3 h 50 min. The total number of start ups was 43. In both cases the biggest running times appeared at night time and shorter ones at day time. This is due to bigger heating need at night time. The differences between running times of bigger pumps could result from different size of houses and different heating system specifications.
Because the current measurements were placed in the common feeder of two different sized heat pumps, running times of smaller pump were difficult to determine. Start up times could still be determined. For the smaller pump of customer 1 time between two start ups varied from 19 min to 2h 10 min and the average time between two start ups was 39 min. The total number of start ups in one week was 255. For the smaller pump of customer 2, time between two start ups varied from 24 min to 1 h 40 min and the average time between two start ups was 30 min. The total number of start ups in one week was 371.
The total number of different heat pump start ups at customer 1 in each hour during one week is shown in figure 7. The figure 7 presents the number of start ups so, that if a smaller pump of customer 1 has started up once every day between 8.00 am and 9.00 am the number for smaller pump of hour 8.00 will be 7. As shown in figure 7, start ups of the smaller heat pump most commonly occurred at evening and day time. At night time, the number of start ups of smaller pump decreased significantly. This happened because more water was used at day time and evenings than at night time. For the bigger pump there was no specific time for start ups, only there was less start ups at morning. This occurred because heat pump ran longer at morning. This was because of bigger heating power need due to decreased outside temperature. Because of this kind of distribution in start up times, there was lower long term flicker severity index at night than day time. The key issue of high short term flicker index was bigger pump start up time and common start ups in the same 10 minute period for both heat pumps of customer 1. Also if there was common start up in the same 10 minute period for heat pumps of customer 2, it raised short term flicker severity index above 1, which is the irritation threshold.
The total number of heat pump start ups at customer 2 in each hour during one week is shown in figure 8. There were more start ups of smaller pump for customer 2 than customer 1. The number of start ups remained quite high all day. There were only little less start ups at night time than at day time. For the bigger heat pump there were only a few start ups during whole week. This was because of long running times of the bigger heat pump. The long term flicker severity index got higher values at evening than at night time because of this kind of distribution of start up times of different heat pumps. For high short term flicker severity index the key issues were the same for customer 2 as for customer 1. The start up of bigger heat pump at customer 2 meant higher short term flicker severity index at customer 2 and if there was start up of both bigger and smaller heat pumps at customer 2 in the same 10 minute period short term flicker severity index got even higher. Also if there was start up of both heat pumps at customer 1 short term flicker severity index increased above 1. In Table 3 short term flicker severity indices are summarized at different customer ends when different heat pumps started up. Short term flicker severity indices are average 10 minute values of events mentioned in table 3. Table 3 shows previously mentioned cross disturbance from start up of heat pump of one customer to the short term flicker severity index of the other customer. Table 3 also shows the greater effect of the bigger heat pump start up to the short term flicker severity index.
Table 3. Average values of short term flicker severity index at different customer end when different heat pump is or pumps are started up
Even though there were sometimes relatively high short term flicker indices the long term flicker index never exceeded the level Plt = 1 during one week. The highest long term flicker severity index was 0.99 at customer 1. The standard EN 50160 defines the threshold level to the flicker so that “under normal operating conditions, in any period of one week the long term flicker severity caused by voltage fluctuation should be Plt ≤ 1 for 95 % of the time”. This also means that threshold defined in the standard was not exceeded.
Despite the fact that flicker did not exceed the threshold level of standard EN 50160, flicker from heat pump start ups was clearly visible. The start ups of heat pumps were easily seen from lighting. Some customers can be irritated from this kind of rapid voltage changes. Now in the distribution utility point of view it is easy to say, that no problems occurred and case is closed. But in the customers point of view there might be flicker problems so the result “there cannot be flicker” is not a good answer from customer service point of view. In this case the result was that the supplier of all these heat pumps will install softstarters to heat pumps.
In this paper there were examinations about power quality problems caused by loads that are coming more and more common. The examination is based on practical case measurements made in real distribution network in Finland. Two groups of loads were selected to this examination: wood splitters and heat pumps.
Wood splitter, which nominal power of induction motor was 2.2 kW and which was single phase device, was selected to this examination. This wood splitter was operated in different locations in one rural area low voltage network in Finland. In this network calculated short circuit currents varied from 1.4 kA to 146 A. Measurement places were selected so that measured short circuit currents were 148 A, 275 A and 350 A at customer supply terminals. In these places the wood splitter caused lots of flicker problems because of a high starting current and because of operation logic which caused lots of start ups of induction motor of the wood splitter. Flicker problems occurred in all three phases because of the star point displacement due to operation of a single phase device. Flicker problems caused by the wood splitter extended also to nearby customers along the same feeder. One way to prevent these flicker problems is to accept in the market only wood splitters of which piston movements are controlled with hydraulic control while the induction motor is running continuously. In such a case there will be less start ups of the induction motor.
Power quality problems caused by heat pumps were studied in urban area network in Tampere. In urban area network there was four big heat pumps installed into two terrace houses. Here start ups of heat pumps caused short term flicker severity index increasing over irritation threshold 1. Also start ups of heat pumps in one terrace house increased the short term flicker index over 1 at the other terrace house so there was cross disturbance from one terrace house to the other. Even though the short term flicker index was sometimes over 1 the long term flicker index was always under 1 so from the standard point of view there was no flicker problem. Despite this every start up of heat pumps could easily be seen from lighting so someone could feel this to be irritating. In this case the result was that the supplier of the pumps will install softstarters to all heat pumps.
 European lamps become more energy efficient. [WWW]. [Cited 7/4/2010]. Available
 Sulpu. Finnish heat pump association. [WWW]. [Cited 7/4/2010]. Available at:
 European Heat Pump Statistics Outlook 2008. European heat pump association. [WWW].
[Cited 7/4/2010]. Available at:
 Finnish government notice. [WWW]. [Cited 31.5.2010]. Available at:
 The great price comparison of air heat pumps. [WWW]. [Cited 31.5.2010]. Available at:
 Official Journal of the European Union- Commission Regulation (EC) No 244/2009
 Math H. J. Bollen. Understanding Power Quality Problems. Voltage Sags and
Interruptions. Gothenburg 2000. IEEE Press. p. 543
 EN 50160. European Standard. Voltage characteristics of electricity supplied by public
distribution networks. 2004.
 Suurinkeroinen, Seppo. Presentation. Luosto Power Quality Seminar 2007.
 Matsch, L. Morgan , D. 1987. Electromagnetic and electromechanical machines. Third
Edition. New York. John Wiley & Sons. 566 p.