Roman Targosz and Jonathan Manson

　　Apart from specifying the real PQ cost of wastage, respondents also defined those hypothetical costs that would be potential losses and risk avoided by power systems that had been immunized against the PQ disturbances under review.

　　The LPQI survey is based on 11 individual cases per complete interview. The subsequent regression analysis was performed to estimate PQ cost across those sectors that offered a convergence of four specific indices. These indices initially were: employment, energy consumption, contractual power, annual turnover.

　　After refining them, the study concluded that annual turnover is a key indicator for a regression model (see Table 18.7 and Table 18.8 below).

　　To arrive at a statistically significant and acceptable model the survey sample was divided into two subsamples: industry and services. The banking sector was excluded

　　because of its anomalous size and structure.

　　In the industry model, analysis shows that for the industry sector the estimation of how much wastage is caused by poor PQ is 4% of annual turnover.

　　In the services model, the estimation of wastage caused by poor PQ is 0.1419% of annual turnover.

　　Statistical bias is a real danger in research like this, especially in terms of how representative the study is of the target universe. This was resolved once the random and statistically based samples were checked. The regression analysis in the LPQI survey project proved that the samples and models were large and good enough to conclude that the variation explained by the model was not due to chance and that the relationship between the model and the dependent variable, which is annual PQ cost, was very strong.

　　The charts in Figure 18.3 present the cost extrapolations of wastage caused by the range of PQ phenomena throughout the sectors investigated in EU-25: PQ cost is characterized by disturbance type (absolute value in Ebn and % value of total cost) and cost components.

　　The cost of wastage caused by poor PQ for EU-25 according to this analysis exceeds E150bn. Industry accounts for over 90% of this wastage. That the proportion of these total PQ costs/losses accounted for relatively by services is possibly explained by cost underestimations by service sector organizations that often experience PQ problems in, say, an office environment, where distinguishing between the cause of a given PQ issue and other root causes may be difficult.

　　Furthermore, some service sectors like data centers, which probably experience high PQ costs, are not represented in the survey. Hospitals fit well into the services model and demonstrate slightly higher PQ costs than other service sectors.

　　Dips and short interruptions account for almost 60% of the overall cost to industry and 57% for the total sample.

　　Figure 18.3 Extrapolation of PQ cost to EU economy in LPQI surveyed sectors [11(Reproduced from the 2007 Leonardo Power Quality Initiative Survey, R. Targosz)

　　This extrapolation corresponds well with those levels indicated by the CEIDS survey in 2000 which reports between $119 and 188bn as the cost of poor PQ-generated wastage in the USA with 4% of companies reporting annual costs of 10% or more of annual revenue and 9% reporting costs of between 1 and 9.99 %.

　　The LPQI survey shows that the economic impact of inadequate PQ costs industry some 4% of turnover and services some 0.15 %. Whilst these values are extrapolations based on the sample interviewed (42 industrial companies and 21 service companies), it can be said with confidence that significant differences exist between the two.

　　Among industrial companies the highest values occur in typical continuous manufacturing industries and lower values in the metallurgical, food and beverages, and general manufacturing sectors.

　　Within the service sector hospitals are significantly higher than other groups.

　　Table 18.7 and Table 18.8 present the statistics relating to different PQ cost frequencies, grouped by industry and services.

　　The three industry histograms presented in Figure 18.4 show the distribution of frequencies for different types of PQ cost indices. They show that in the case of PQ cost per turnover, the frequency is closer to a normal distribution curve and the ratio of mean to standard deviation is far lower than in the case of other indices

　　Also, when comparing other statistics, especially variance, it is clear that the most accurate model would be based on PQ cost per company turnover.

　　When analyzing other indices, it is also apparent that the more appropriate representation is values that are closer to the median rather than the mean.

　　The general structure of PQ cost for each sector is presented in Figure 18.5.

　　It is noticeable that in typical continuous manufacturing sectors the losses incurred by lost work-in-progress (WIP) is quite significant and responsible for about one-third of the PQ costs recorded.

　　The slowing down of processes, which sometimes integrates WIP, and labor cost where these are not visible as independent components, is understandably even more significant.

　　In other sectors the situation is less clear with either labor cost or equipment-related costs being the most important source of economic losses.

　　Finally, in relation to public services like hotels and the retail sector, PQ impact is measured as slowing down their business activities, in terms of revenues that are irrevocably lost.

　　Figure 18.6, Figure 18.7, Figure 18.8, Figure 18.9 and Figure 18.10 respectively present PQ cost structures for five major groups of PQ disturbances – dips, short and long interruptions, harmonics, and surges and transients.

　　Figure 18.10 PQ cost components per disturbance type [11] (Reproduced from the 2007 Leonardo Power Quality Initiative Survey, R. Targosz)

　　The specific observations are:

　　• Voltage dips – WIP accounts for almost 50% of PQ cost and the largest single source of PQ-caused economic losses, with process slowdown accounting for a further 30 %.

　　• Short interruptions – the cost structure is similar but costs relating to equipment failure/premature ageing are more significant. This is in part explained by the influence of the semiconductor sector. This sector claims to be fully immunized against dips; however, evidence suggests a lack of immunity to short interruptions, in which the category of equipment-related cost dominates.

　　• Long interruptions – the importance of labor costs increases, reaching some 20% on average. This is twice as great as the equivalent figure for voltage dips and some 40% higher than with short interruptions. In addition there are instances where some of the other cost categories take on greater importance and these tend to be related to the long-term economic consequences created by penalties (commercial or statutory), loss of brand equity or the need for unanticipated business investment to regain lost sales/market share.

　　• Harmonics – process slowdown generates almost two-thirds of all harmonics-related costs. Equipment-related costs represent about 25% of these harmonics costs. However, only 8 out of 62 companies specified PQ cost of harmonics that related to additional energy losses. This cost in total is E186 000 and represents about 1% of the total cost of harmonics in the sample.

　　• Surges and transients – production outage is again a major source of economic losses and is responsible for two-thirds of total PQ cost and consequential losses. An interesting observation from one of the continuous manufacturing companies (see Figure 18.10 – the bottom chart) was that 90% of transient/surge cost was claimed successfully (presumably to the electricity supplier).

　　The breakdown of costs by different PQ disturbances is presented in Figure 18.11.

　　On average the absolute share of impacts (before sector grouping and extrapolation) of the six categories of disturbances taken from the total survey sample is as follows:

　　• Voltage dips 23.6%

　　• Short interruptions 18.8%

　　• Long interruptions 12.5%

　　• Harmonics 5.4%

　　• Surges and transients 29%

　　• Other 10.7%

　　These shares can be further summarized as:

　　• Voltage dips are the most important source of impacts in the continuous manufacturing sector.

　　• Short interruptions are most significant for food, metallurgy and newspaper publishing.

　　• Long interruptions are most costly for hotels and other public service sectors.

　　• Surges and transients are most destructive for the telecommunications sector and, perhaps surprisingly, for the pharmaceutical sector.