By JJ Lopes Morgado

INTRODUCTION
With the prospect of long lasting global financial and economic turbulences for the next two years and with the renewal of the concession contract for the coming years, Companhia De Electricidade De Macau (CEM) has to find strategies and action plans to cope with the coming challenges. One of these strategies is the introduction of Automatic Meter Management (AMM). The C&I customers’ AMM was already successfully implemented on July 2001 and CEM is about to complete an AMM pilot project of 1,000 residential and commercial customers. In this article we are going to talk about the residential and commercial AMM (RCAMM), which will be referred to as simply AMM hereafter. As we know, for this type of project, usually there are important drivers in terms of cost justification, and less important ones. The Macau environment also includes both positive and negative factors or drivers which will affect the viability of our AMM. Most of the drivers found in other AMM projects and countries are not strong enough in Macau. A long list of issues such as safety, high manual meter reading costs, a high level of meter fraud, low density areas, a low level of daily meter reads per meter reader, a high number of ‘hard to read’ locations, a high rate of meter re-reads, an inefficient read to bill time, a high level of occupancy turnover, etc do not work in Macau at all. But if we can prove that some drivers have an important role and are able to be confirmed (for instance, the non-technical losses, the prepayment solution, the usage of a reliable Power Line Communication (PLC) network at low cost and some other value-added services and technologies), then we can surely justify the Macau AMM in terms of costs and secure a shorter payback period for the project.

The AMM pilot project was split into three parts and delivered to three suppliers (each one integrating 330 customers) utilising different solutions and technologies. The three solutions have the same architecture that is considered the most suitable for Macau. We believe the solutions we are working on are cost effective and include many features considered essential by CEM, to meet our needs for some years to come. One of these solutions is the magic word – PLC (Power Line Communication).

THE PLC EXPERIENCE IN CEM
Until recently, most of the AMM systems and homes’ and buildings automation was realised through systems which need a special transmission medium, such as s pair of twisted wires, coaxial cable or optical fibre. Recent technological developments led to power line medium equipments which send and receive information with some reliability. The main advantage of a PLC system is that the physical medium is already installed, making it an attractive alternative in all buildings without pre-routed data infrastructure, like historical buildings or brief local data networks. Due to this existing infrastructure, digital communication over the power lines has become an excellent opportunity for the utilities to implement new services. Digital communications on the low-voltage electrical grid is relevant to a number of industrial actors, such as electricity, gas, heating and water distributors. Typical applications using PLC include Outage Management Systems (OMS), Automatic Meter Management (AMM/AMI) systems, Smart Grids, GIS systems, remote control tasks, load management systems, tariff switching systems, etc.

THE CHARACTERISATION OF THE PLC CHANNEL
The power lines were originally designed for distribution of 220/110 power at 50-60 Hz and not for communication purposes. To make matters worse, the power line is one of the most electronically contaminated environments in the world and the signal attenuation at the frequencies of interest is fairly high, which makes it more challenging to achieve dependable results. As a consequence, its properties as a communication channel are still not fully understood. Using this medium for narrowband communications at higher frequency bands presents many technical problems.

The low-voltage power line network is made of a variety of wiring types, connected in almost random ways (which has a strong effect on impedance mismatch). In Macau this phenomenon has a significant impact. In addition to that, very different types of devices are part of the LV network (electricity meters, fuses, etc.) and a large variety and quantity of appliances, equipment and devices can be connected in any point (air conditioners, washing machines, TV sets, etc.) and have a significant impact on the quality of the PLC channel. In fact, in this aspect, Macau has the worst case scenario.

This type of network has a complex frequency response (amplitude and phase) that changes both in time and frequency. At some frequencies, the received signal can be very strong, while in other frequencies the received signal might be too weak to be usable for reliable data transmission.

An important property of the electrical signal is its bandwidth (the width of the frequency interval around the carrier frequency that is occupied by the electrical signal), and there exists a close (proportional) relationship between bandwidth and bit rate of the communication. Another aspect of fundamental importance that affects the bit rate is the quality of the power line communication channel. Hence, not only the available bandwidth determines the bit rate that can be used, but the quality of the channel also does. However, the available bandwidth is in general the single most important parameter for making high bit rate applications possible.

There are several factors that affect the quality of the power line communication channel and define its characterisation. They are attenuation, noise, and impedance. These vary not only with the location, but also over time, allowing the establishment of a time relation with the day period. Thus, the PLC system must be capable of adapting to inconstant power line characteristics and of achieving the higher transmission rate possible in such conditions.

  • The attenuation
    For low frequencies the attenuation is closely related to the distance between the emitter and the receiver. In high frequencies, due to reflections by mismatched impedances, the signal attenuation in the medium is almost not dependent on the distance between the emitter and the receiver. The propagation and attenuation of high frequency carrier signal are very dependent upon the power line structure, current variation, and loads, among other factors. Also, the attenuation between different network phases is similar to the verified for the same phase for longer distances.
  • PLC1

    Figure 1: Noise Characterisation

    The noise
    The noise on power lines is mainly caused by electrical appliances connected to these lines. So the statistical behaviour of this man-made noise is quite different from that of stationary white Gaussian noise and its characteristics may change in very short time periods. Therefore a model which can describe the statistics of the instantaneous value of the noise is difficult to achieve. A commonly accepted model is the one referred in figure 1.

For frequencies up to 145 kHz, where the noise assumes higher values, the noise is essentially due to:

  1. Coloured background noise (produced by electrical motors, microwave ovens and light dimmers)
  2. Narrowband noise (caused by radio AM broadcast signals ingress and TVs horizontal synchronism frequency)
  3. Asynchronous impulsive noise (produced by on and off switching events).
  • The Impedance
    A power line has highly variable impedance depending on several factors such as its configuration (star connection, ring connection) or the number of appliances linked. On the other hand, impedance is practically constant in time but substantially changes in frequency, thus indicating a clearly inductive behaviour (jWL).
    The impedance of the residential power circuits increases with frequency and is in the range of about 5 to 32 O at 100 kHz. Some authors also report a characteristic impedance varying from 70 to 100 O for different types of wires.
    The impedance is determined by two parameters: the loads connected to the network and the impedance of the distribution transformer. Recently a third element has influenced in a relevant way the impedance of a power line, in particular in the residential network. It is represented by the EMI filters mounted in the last generation of home appliances (refrigerators, washing machines, television sets, hifis). Wiring seems to have a relatively small effect on impedance.

THE PROTOCOLS (PLC COMMUNICATION IN THE ‘LAST MILE’)
As previously stated, the main advantage of power line communications is the fact that the physical network is already installed, making it a very attractive option. Given the physical drawbacks of the network, it becomes necessary to install repeaters of some sort if the distances are appreciable and bridges over distribution transformers if there is a need to go over that boundary.

Another aspect that needs to be considered is the difference in standardisation according to the part of the world where the technology is to be set. The most important parameters specified by the different standards are the maximum transmitted power and the allowable bandwidth. The restriction is imposed to limit the interference with other telecommunication services and maybe as a measure to avoid further spectrum contamination.

It is important to note the frequency spectrum limitation imposed by the regulatory agencies in the two most important markets: North America and Europe. While in Europe, power line communication is restricted to operate in the frequency spectrum ranging from 85 to 150 kHz; in North America the spectrum is wider (540 kHz).

The stricter limitation put into effect by Europe’s CENELEC EN50065-1 and the narrower bandwidth allowed for power line communications may preclude the use of one of the standards in some regions, as will be described later.

The FFC Standard (American standard similar to CENELEC) allows a wider band, but according to some authors can disturb LW radio transmission.

THE RF EMISSIONS IN POWER LINES
Electrical transmission lines were designed to conduct 50 or 60-Hz power from point to point. At those frequencies, power lines are excellent transmission lines and little of that 50 or 60-Hz power is radiated. Besides the PLC based AMM applications, the electric-utility industry also uses these lines at frequencies below 500 kHz to transmit Power Line Carrier signals to control utility equipment. These lines, however, were not designed to carry radio frequency (RF) energy. As the frequency of carrier-current signals conducted on power lines is raised, the amount of signal radiated from the line increases rapidly. For a given signal level, the radiated emissions from power lines will increase by tens of dB as the frequency is increased from LF through HF. Just for curiosity, at HF, power line wiring makes a fair to excellent antenna, similar in gain and pattern to the antennas used by licensed radio services.

Regarding the RF emissions it is commonly accepted that:

  • Above 30MHz the radiated emissions are predominant.
  • Below 30MHZ the conducted emissions are predominant.

THE CEM CASE
In CEM trials, as far as PHY (1st layer) and DLL (2nd layer) layers of the OSI 7 layers model is concerned, both CENELEC and FCC protocols were used by the selected suppliers for the AMM pilot project. CENELEC protocol modems are installed in customer substations (PT) with transformers of 1000 and 1500 kVA (Low Load Transformers sites – LLT) and FCC protocol modems are installed in PT’s with transformers of 1600 kVA (Medium Load Transformers sites – MLT). Also modems with CENELEC protocol, used in LLT sites, are from 2003 (first generation) while FCC protocol modems, used in MLT sites, are from 2008 (last generation).

We found that LLT sites have on average a lower noise level than MLT sites due to the smaller transformers installed. The results are shown (Figure 2 and 3):

  • LLT sites showed an average noise level below 72 dBuV.
  • MLT sites showed an average noise level of 88 dBuV.

The measurements were taken on April and October of 2008. We believe that in July and August these measurements are slightly higher.

For both protocol types, CENELEC and FCC, we had to use repeaters (not many) for the most difficult cases/situations. The repeaters we used are simply PLC modems that behave as repeaters. Our PLC modems are installed at the meter room level in each floor.

The average success transfer rate on a daily basis achieved by both protocols is approximately:

  • for CENELEC protocol: 98 %;
  • for FCC protocol: 100 %
PLC2

Figure 2: Power Line Noise Spectrum – LLT sites –
Average Noise Value = 72 dBuV (-35 dBm)

WHY IS NOISE LEVEL SO HIGH IN MACAU?
Several worldwide PLC experts that conducted trial tests in Macau for the past five years have stated that the Macau LV powerlines network situation is much more difficult for communications than in Europe and Japan. The average noise level found in Japan and Europe is around 50-60 dBuV for high season and daily peak hours, which is viable for digital communications. In Japan, for instance, only in-house (indoor) PLC is allowed. Japanese regulation requires suppressing the noise from the PLC installation as low as the ambient noise, at the distance of 10m from the house, to protect the shortwave radio users in neighbouring houses. However, even with tight regulations, the reality is that the PLC noise exceeds the ambient noise level and often exceeds more than 10dB and causes serious interference with the radio reception.

In Macau for the trial sites we have selected, with around 300 customers in each transformer (which is not the worst case; we have many transformers with more than 1000 customers), the average noise level is around 80 to 100 dBuV during the summer season and at the peak hour (worst-case scenario).

This situation can be explained by the following factors:

  • Many of the electrical appliances we have in Macau do not follow the international standards and do not have the EMI filters mounted in the power cable (most of them)
  • The Neutral’s Operating System in the Low Voltage electrical network we have in Macau (which is a combination of TT and TN-S systems) may also influence this situation
  • The old power cables we have in some buildings are not compliant with the standards anymore
  • The heavy machinery on one side and thousands of nonstandard electronic devices, on the other hand, installed inside the buildings also have a great impact in the current situation.
PLC3

Figure 3: Power Line Noise Spectrum – MLT sites –
Average Noise Value = 88 dBuV

Due to the above factors, Macau is facing some difficulties in terms of a straight solution for an AMM project based on PLC communications. We believe that FCC protocol may have a more robust and reliable behaviour than CENELEC protocol for the Macau LV electrical networks environment, but it is still a bit early to state that. During 2009 further tests will confirm this assumption. We also believe that some neighbouring regions like Hong Kong and even Singapore face similar problems.

Conclusions
Based on our five years’ experience in the Power Line Communications (PLC) technology field, we think a PLC-based AMM system is technically viable in Macau. Even though some research companies believe PLC is destined to become a technology curiosity for companies with spare cash, PLC technology has been viable for some years, as demonstrated by several commercial products available today. Advances in signal processing and semiconductor development have made this concept a reality. These technologies obtain data transfer speeds of up to 10 KBPS (with some others being announced with speeds above 100 KBPS) which is good enough having in mind all functionalities required for this type of AMM system.

The assessment of available data through the CEM trials is not completed. It is expected to be finished by mid 2009. We believe that a PLC solution for AMM is not a constraint anymore in Macau. Moreover, as we have said before, if we can prove that some drivers can be confirmed (Non-technical losses, Prepayment and some value-added services), then we can surely justify the Macau AMM in terms of costs and secure a shorter payback period for the project.

An additional comment regarding non-technical losses as an eventual driver for Macau AMM is warranted: We can say that in the last decade, the total CEM system losses were 4.5 to 5 percent of the total energy generated. Of this, it is estimated that 4 percent was due to technical losses and 1 percent due to non-technical losses. Should the AMM trial results show non-technical losses have been reduced by 50 percent, or to 0.5 percent of the total energy supplied to the 1,000 trial participants, and if this reduction is consistent over the total CEM customer base of 220,000 consumers, then a cost saving of over US$ one million per year will be realised. This, together with the expected savings to be produced by other benefits/drivers, will provide a strong case to justify the expansion of the trial AMM system at CEM.