Assessment of economic impact of electricity supply interruptions in the Sri Lanka industrial sector

Energy Conversion and Management 45 (2004) 235–247www.elsevier.com/locate/enconman

Assessment of economic impact of electricity supplyinterruptions in the Sri Lanka industrial sector

Priyantha D.C. Wijayatunga a,*, M.S. Jayalath b

a Department of Electrical Engineering, Centre for Energy Studies, University of Moratuwa,

10400 Moratuwa, Sri Lankab NEXANT SARI/Energy, A Bectel Affiliated Company, Hotel Lanka Oberoi, Galle Road, Colombo 3, Sri Lanka

Received 20 December 2002; accepted 24 May 2003

Abstract

This paper presents the outcome of the Sri Lanka case study on assessing the economic impact of power

interruptions on industry in the South Asia region, comprising the countries of Sri Lanka, Nepal, Ban-

gladesh and India. The technical assessment evaluates the cost to the country�s economy in terms of the

industrial loss due to supply interruptions and environmental impacts from standby generation used to

supplement the power requirements of the industrial sector.The study found that the main economic impact of the power interruptions, both planned and un-

planned, is the loss of output in the industrial sector. In a typical year of power shortages, such as 2001,

arising from a deficit in generation capacity, these losses can be as high as approximately US$ 81 million a

year, which is approximately 0.65% of the country�s gross domestic product (GDP). Also, the economic

impact due to unplanned outages can be around US$ 45 million (0.3% of GDP) in a typical year. On

average, these values for planned and unplanned outages are US$ 0.66 and US$ 1.08 per kWh of energy

loss, respectively.

It is also observed that 92% of the sampled industries have standby generation facilities to satisfy either, infull or partially, their own power requirements, which produced approximately 146 GWh of energy in 2001.

The serious economic and environmental impacts of power interruptions, both planned and unplanned,

underlines the importance of timely implementation of the long term least cost generation expansion plan

and proper maintenance of transmission and distribution networks to ensure their high reliability. There-

fore, it is clear that the utility needs to take immediate steps to improve its supply reliability in order to retain

consumers and justify the existence of a centralised generation facility.

� 2003 Elsevier Ltd. All rights reserved.

Keywords: Unserved energy; Cost of outages; Economic impact

*Corresponding author. Tel.: +94-1-650279; fax: +94-1-650622.

E-mail addresses: [email protected] (P.D.C. Wijayatunga), [email protected] (M.S. Jayalath).

0196-8904/$ - see front matter � 2003 Elsevier Ltd. All rights reserved.

doi:10.1016/S0196-8904(03)00132-8

mail to: [email protected]

236 P.D.C. Wijayatunga, M.S. Jayalath / Energy Conversion and Management 45 (2004) 235–247

1. Introduction

The power supply industry in Sri Lanka is a vertically integrated system, mainly owned by thegovernment with the exception of a few generating stations owned and operated by the privatesector. State owned Ceylon Electricity Board (CEB) owns and operates a large majority of thegenerating stations and the transmission system. While the CEB supplies about 90% of theconsumers connected to the distribution system, Lanka Electricity Company (LECO), whosemajority stock is held by the CEB, supplies the remaining 10% of the consumers [2].

1.1. Electricity demand

The overall annual electricity demand grew from 823 GWh in 1972 to 5416 GWh in 2000 at anannual compound growth rate of about 7%. The annual growth rate is expected to be about 8–9%in the next decade. Of the total electricity consumption in 2000, 38% was attributed to the in-dustrial sector, while the household and the commercial sector consumptions were 40% and 20%,respectively [1].

The country wide electrification level stands at approximately 60% of the households or about 2.9millionhouses in the countrywitha relatively largepercentageof ruralhouseholdsnotbeingconnectedto the national grid. There is a vast disparity of electrification levels across the countrywith about 85%of the households electrified in the urban areas, while it is only about 45% in the rural areas [1].

1.2. Electricity generation

The use of hydroelectricity in a particular year entirely depends on the availability of effectiveinstalled capacity and rainfall in the catchment areas. CEB�s generation plan opts for the usage ofoil fired thermal plants only if there is still an energy requirement after all hydroresources havebeen utilised.

The total hydroelectricity generation capacity of about 1150 MW in existing power stationscontributed about 3197 GWh to the electricity supply in 2000. Though there is an estimatedmaximum potential of about 2000 MW hydropower, exploitation of the remaining potential isconstrained by economic factors. Thermal generation plants with an installed total capacity of 685MW supplied 3486 GWh in 2000. This includes utility owned generation, independent powerproduction and captive generation. Additionally, a 3 MW pilot wind turbine plant in the southcoast of Sri Lanka produced 3.5 GWh in the same year [1].

Expansion of the hydroelectric system during the period 2002–2017 is limited to an installedcapacity of 220 MW, giving an annual average energy output of about 833 GWh. This impliesthat there will be a significant difference between the demand for power and hydropower output,which will need to be bridged by thermal plants, which are either oil or coal based according to thepresent generation plan [1].

1.3. Cost of unserved energy

The cost of energy not served is an important parameter used in generation expansion planning,and it will finally determine the system reserve requirements. This value is of special significance in

P.D.C. Wijayatunga, M.S. Jayalath / Energy Conversion and Management 45 (2004) 235–247 237

the country�s important sectors such the industrial sector, which is heavily dependent on theelectricity supply reliability. There are several approaches to estimating unserved energy cost for agiven power system [8]. The three most commonly used techniques are [6] estimation of

(a) the value of production loss for each unit of power outage,(b) cost of alternative or back up generation,(c) willingness to pay for reliable and uninterrupted electricity supply.

The study presented in this paper uses the method given in (a) for the detailed estimation of thecost of unserved energy in the Sri Lanka industrial sector. A study conducted in India indicatesthat this figure is approximately US$ 0.54 per kWh and US$ 0.16 per kWh in the states ofKarnataka and Haryana, respectively [6]. Also, a similar study in Bangladesh estimates this valueto be approximately US$ 0.8 per kWh [7]. Not only do these values differ from system to system,but also they vary widely depending on the duration and the time of interruption [8–12].

2. Electricity–economy correlation

Fig. 1 shows the strong correlation between power shortages (effective demand satisfied) and theeconomic activity in Sri Lanka for the period 1977–2000. When the economy of the country wasseriously affected due to the social uprising in 1989/90, the electricity demand shrank drastically,while the power shortages in 1996 resulted in a substantial reduction in the economic growth rate.The correlation coefficient between the gross domestic product (GDP) and electricity growth ratesis 71.6% based on historical data. Therefore, it is understood that any unfavourable gap betweenthe demand and supply of electricity in the country adversely affects the country�s economy [1].

3. Power shortages

Since the early 1990s, Sri Lanka has been experiencing power shortages from time to time,mainly because of a shortage of hydropower resulting from severe drought conditions. While

0%

2%

4%

6%

8%

10%

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

Year

Gro

wth

Rat

e (G

DP)

-10%

-4%

2%

8%

14%

20%

Gro

wth

Rat

e (E

lect

rici

ty)

GDP Electricity

Source: CEB Generation Planning Division

Fig. 1. Correlation between electricity demand and economic activity.


these droughts have been occurring once every four to five years, the intensity of the resultingpower shortages has been increasing in recent years because of the ever growing electricity de-mand coupled with inadequate thermal capacity additions. In particular, during the last decade,the implementation of capacity additions, as proposed in the long term generation expansion planof the CEB, has been always delayed. The worst power shortages occurred in 1996 and in 2001when rotating power interruptions (load shedding) were introduced to reduce the daily demand inorder to facilitate proper grid operation. Approximately 300 h of load shedding had already beenplanned by the CEB for the year 2001 to cope with the imbalance in the power supply–demandequation.

These power shortages, occurring now almost on a regular basis, emphasise the need to havealternative forms of power supplies, particularly for meeting the needs of the commercial andindustrial sectors where the effects of such shortages are likely to be substantial in economic terms.Most of the large and medium scale installations in these sectors, therefore, have opted forstandby generation (SBG) to satisfy, at least partially, the essential electricity requirements ofindividual installations under such circumstances.

4. Study objective

The main objective of this study is to determine the economic cost of both planned and un-planned power supply interruptions in the industrial sector in Sri Lanka. This is a component of alarger regional study where the same exercise is replicated in Nepal and Bangladesh.

Further, the penetration of standby generation facilities and their incremental impact of en-vironmental emissions due to grid supply outages are also examined during the study.

5. Methodology

This study is based on a survey conducted with a selected sample of industries representingdifferent industry sector sub-groups. The population was characterized and the samples wereselected based on the considerations listed in Section 5.4. Using the developed questionnaire, datacollection was conducted through personal visits. Data that was obtained from the field researchwas then centrally archived for analyses and reporting.

5.1. Population characterization

Using the standard industry classification published by the Central Bank of Sri Lanka (Cate-gories 1–9 in Table 1), the contributors to the GDP were listed in the order of their absolute andpercentage contribution to the GDP. Noting that some categories contributed very little to theGDP and that there were only a handful of those contributors, the study focussed on thosecategories that contributed more than 4% to the total GDP. In addition to these categories, theTea, Rubber and Coconut Processing Industries and Hotels were included in the study due to thewidespread nature of their operations in the country. The GDP contribution of each of thesecategories in year 2000 is shown in Table 1 [3].

Table 1

GDP contribution of each industry category considered

Categories GDP US$

(million)

Percentage

(%)

(1) Food, beverages and tobacco 545 22

(2) Textile, wearing apparel and leather products 772 32

(3) Wood and wood products 17 1

(4) Paper and paper products 31 1

(5) Chemical, petroleum, rubber and plastic products 197 8

(6) Non-metallic mineral products 158 7

(7) Basic metal products 11 0.4

(8) Fabricated metal products, machinery, transport equipment 86 4

(9) Products not specified elsewhere 44 2

(10) Tea 158 7

(11) Rubber 23 1

(12) Coconut 133 5

(13) Hotels 250 10

Total 2425 100

Source: Central Bank of Sri Lanka, Annual Report 2000.


5.2. Sample

Industrial categories that did not have significant contributions to the annual GDP (less than4% of the total contribution of the population considered) were removed from the sampledpopulation. This resulted in the removal of the following four categories.

• Wood and wood products (1%)• Paper and paper products (1%)• Basic metal products (0.4%)• Products not specified elsewhere (2%)• Rubber (1%)

Table 2 shows the study categories and the number of industrial installations in each categorythat remained after this initial filtering process. A total of 150 samples were then selected from thisremaining population, consisting of a total of 3074 industrial installations spread over eightdistinct categories and 16 sub-groups.

The number of samples chosen from each category was based on the proportion of the totalcountry-wide number of installations in that category. In order to have a fair representation of thepopulation within each sub-group, the samples were chosen randomly within each group, givingemphasis to the fact that the industrial consumers in each of these categories have varying elec-tricity intensities.

Fig. 2 shows the final number of samples from each industrial category after the filteringprocess. Given the criterion for sample selection from the first five categories, the number ofsamples from the Textile and Leather category became the largest.

Table 2

The industry population used in the study

No. Industry type Industry population

1 Food, beverage and tobacco 321

1.1 Food and other 287

1.2 Liquor 18

1.3 Beverage 8

1.4 Tobacco 8

2 Textile and leather 728

2.1 Apparel 497

2.2 Textile 169

2.3 Leather 62

3 Chemical, petroleum, rubber and plastic products 521

3.1 Chemical 41

3.2 Rubber 138

3.3 Plastic 193

3.4 Pharmaceuticals 147

3.5 Petroleum 2

4 Non-metallic mineral 143

4.1 Diamond 20

4.2 Ceramic 33

4.3 Cement 12

4.4 Building 78

5 Fabricated metal, machinery, transport equipment 349

6 Tea 745

7 Coconut 60

8 Hotels 207

Total 3074


5.3. Questionnaire

Determination of the cost of unserved energy in this study is based on the approach where thecost of direct losses to the industrial consumers due to power outages is examined. In order to gatherall the relevant information, a questionnaire was designed. Power interruption problems wereclassified as momentary, planned and unplanned power outages. The effects of voltage fluctuationsand harmonics were also considered in the design, which finally contained the following.

• General information on the industry installation, such as contact information and category ofindustry.

• Electricity consumption information based on monthly metering data.• Financial information, such as the annual turn over, expenditure on raw materials and average

annual value addition of each industry installation.• Revenue losses due to sudden interruptions (momentary losses), unplanned longer interrup-

tions and planned interruptions, such as imposed power interruptions during the times ofpower shortages.

2439

28

20

17

9

3

10

Food, Beverage & Tobacco Products

Textile, wearing apparel and leatherproducts

Chemical, petroleum, rubber andplastic products

Non-metallic mineral products

Fabricated metal products,machinery and transport equipment

Tea Processing

Coconut Processing

Hotel

Fig. 2. Composition of the selected sample.


• Revenue losses due to voltage variations beyond allowable limits, those due to harmonics inthe electricity supply and those resulting from supply restrictions, such as the ban on usingair-conditioners connected to the grid supply during certain periods of the year.

• General comments given by the industry on the electricity supply.• Technical information associated with any standby generators installed in the industry pre-

mises.

The information to be collected was subdivided into different categories of losses such as thoseassociated with raw material, manpower, production output and standby generator usage forconvenience in data gathering.

5.4. Data collection and data organization

Collected data were processed separately for different types of interruptions and other supplyquality problems, such as voltage variations and harmonics. Within each of these, the contri-bution to losses by raw material, output and manpower was tabulated separately. All the gene-rator data was also processed and tabulated separately.

5.5. Data analysis

Data collected for each category of power interruptions and other power quality problems wereanalysed to obtain the average values and their variances for each category within the sample.This led to the losses for momentary interruptions determined in terms of US$ per each inter-ruption, while the losses due to planned and longer unplanned supply outages were calculated interms of US$ per hour of interruption duration.

In order to estimate economic losses from the collected financial loss data, a conversion factorof 0.76, used by the Ministry of Finance and Planning, was used in the analysis.


When extending the economic values of different types of losses to cover the national scenario,it is important to examine the correlation of the values obtained during the sample survey with thenational figures. The ratio of value addition determined for each industry category with sampleddata against the national GDP contribution reported for the same category is used as an indi-cation to obtain the country wide average losses for each category of industries. This scaling isimportant, since the sample cannot capture all possible types of industries and, hence, give widevariances for the average values.

6. Study results

The results of this study are presented in the form of expected average losses experienced byparticipants in the sample. Based on the structure of the questionnaire, the losses due to loss ofraw material, manpower and output are quantified in terms of monetary values and presented bycategory.

6.1. Average industry losses

Based on the analysis presented in Section 5.5, the weighted average values of losses in differentindustry categories may be calculated (weighted within each main category). These values forindividual categories are given in Table 3.

The values given in Table 3 for momentary interruptions are the losses incurred at the momentof interruption. In the case of planned and longer term unplanned interruptions, the losses be-come a function of time. The national proportions of outage costs in different industry categoriesfor planned and unplanned interruptions are given in Figs. 3 and 4.

In the cases of both planned and unplanned interruptions, the main contribution to the na-tional economic losses comes from the following four categories of industries.

• Food beverages and tobacco products• Textile, wearing apparel and leather products• Chemical, Petroleum and Plastic products and• Fabricated metal products, machinery and transport equipment

These losses, determined for different industry categories, vary within a range. The variation oflosses in major categories of industries from the average loss level with 90% confidence forplanned outages is given in Fig. 5 (100% indicates the average value). It can be seen that thesevariations are considerably high. This is mainly due to the wide range of industry installations interms of the nature of output, electricity intensity and output capacity that one finds within eachof these industry categories. A similar range of variations can be seen for unplanned outages [5].

Planned and unplanned power interruptions affect the industries in different ways due to thedifferent levels of preparedness associated with these interruption types. It can be seen from Fig. 6that the overall planned outages in the industrial sector costs only about 60% of that associatedwith the unplanned outages.

Table 3

Economic losses due to supply interruptions in each industry category

Average losses (US$)

Industry category Momentary

(US$)

Planned

(US$/h)

Unplanned

(US$/h)

1 Food, beverage and tobacco products 302 153 363

1.1 Food and other 211 166 279

1.2 Liquor 3 7 7

1.3 Beverage 139 161 161

1.4 Tobacco products 4363 16 4378

2 Textile, wearing apparel and leather products 122 108 186

2.1 Apparel 121 82 161

2.2 Textile 110 134 199

2.3 Leather 158 248 354

3 Chemical, petroleum, rubber and plastic products 156 83 218

3.1 Chemicals, Paints and Fertilisers 221 316 316

3.2 Rubber 67 57 110

3.3 Plastic and PVC 187 110 288

3.4 Pharmaceuticals, detergent and other 180 5 200

3.5 Petroleum products 0 65 65

4 Non-metallic mineral products 169 108 267

4.1 Diamond processing 51 51 51

4.2 Ceramic products 100 98 172

4.3 Cement 782 582 1636

4.4 Building materials and other 135 54 151

5 Fabricated metal products, machinery and transport

equipment

125 416 429

6 Tea industry 4 22 22

7 Coconut industry 1 75 131

8 Hotel industry 0 26 26

Cost of Planned Outages

Food, beverage and tobaccoproducts

Textile, wearing apparel andleather products

Chemical, petroleum, rubberand plastic products

Non-metallic mineral products

Fabricated metal products,machinery and transportequipment

Tea Industry

Coconut Industry

Hotel Industry

41% 5%

22%

12% 4%

1% 1% 14%

Fig. 3. Proportions of cost of planned outages.


Cost of Unplanned Outages

Food, beverage and tobacco products

Textile, wearing apparel and leatherproductsChemical, petroleum, rubber andplastic productsNon-metallic mineral products

Fabricated metal products, machineryand transport equipmentTea Industry

Coconut Industry

Hotel Industry

20% 3% 1% 1% 20%

23%

19%7%

Fig. 4. Proportions of cost of unplanned outages.

0%

50%

100%

150%

200%

250%

300%

Food

,be

vera

gean

dto

bacc

oT

exti

le,

wea

ring

appa

rel

and

Che

mic

al,

petr

oleu

m,

rubb

er a

ndpl

astic

Non

-m

etal

lic

min

eral

prod

ucts

Fabr

icat

edm

etal

prod

ucts

,m

achi

nery

Tea

Indu

stry

Coc

onut

Indu

stry

Hot

elIn

dust

ry

Var

iati

on (%

of a

vera

ge)

Fig. 5. Variation of countrywide expected values of losses under planned interruptions.

Planned

Unplanned

US$ 0.36 million per hrUS$ 0.58 million per hr

Fig. 6. Comparison of cost of planned and unplanned outages.


6.2. National economic losses

Planned interruptions can be due to scheduled regular maintenance in the power system or maybe due to power shortages during times of drought, which drastically limits hydropower input tothe generation system in Sri Lanka. During 2001, planned power interruptions due to powershortages amounted to approximately 300 h up to early December. The study estimated theeconomic loss associated with this 300 h of power interruptions to be approximately in the rangeof US$ 47–117 million (90% confidence level), which is 0.4–0.9% of the country�s GDP in thatyear, which was US$ 12.5 billion.


When the interruptions are planned and consumers are informed well in advance, the losses canbe minimised, particularly the raw material and manpower losses experienced in some industries.Unplanned interruptions are mainly due to unexpected failures in various parts of the powersystem or due to the utility�s failure to inform the consumers in advance of a planned interruption.Generally, these interruptions do not last long and the total period of such losses during a givenyear is considerably lower in comparison to typical planned outages during the periods of powershortages. This was estimated to be about 100 h in the year 2000/2001. The estimated economicloss associated with these interruptions is about US$ 30–58 million a year. In GDP terms, it isapproximately 0.25–0.46%.

The economic losses due to planned and unplanned power interruptions can also be expressedin other forms. One of the commonly used forms is the economic loss per unit of supply loss (US$per kWh), identified as the cost of unserved energy. The study reveals the average cost of unservedenergy for the Sri Lanka system is US$ 0.66 per kWh in the case of planned interruptions, while itis US$ 1.08 per kWh if the interruptions are unplanned.

6.3. Standby generation (SBG)

With the availability of continuous main grid supply in question, many industries have investedin standby generators. Approximately 92% of the industries sampled have standby generators.While standby generators accomplish their main goal of supplying uninterrupted power, theirrelatively low operating efficiency and, hence, increased pollution per unit of energy output aremajor concerns. Based on current usage practices, the study has determined that the use ofstandby generators results in approximately 7% incremental environmental emissions whencompared against utility generation [4].

Based on current usage and penetration levels, a total of about 300 h of power interruptionsannually, it is expected that the standby generation can be as high as 160 GWh. It is to be notedthat the annual electricity generation by the utility was about 7000 GWh in 2001. The incrementalemissions are estimated to be 7400 tons of CO2, 9.5 tons of CO and 10.9 tons of other particulatematter [4].

6.4. Voltage variations and supply harmonics

No industry reported any financial losses associated with the supply harmonics, while only 16%of the installations in the sample indicated possible losses in the event of voltage variations.However, none of these installations had experienced any serious voltage problems during theiroperations.

7. Conclusions

This study has clearly shown that electricity supply interruptions in Sri Lanka can cause sig-nificant losses in the industrial sector, and hence, they seriously impact the national GDP. This


impact can be as high as about 0.9% of the GDP in the country where it experiences only around5% annual GDP growth on average.

While quantifying planned and unplanned outages, this study shows that losses due to un-planned interruptions, inclusive of momentary interruptions, are approximately 1.6 times thosedue to planned outages. This means that all efforts must be taken to reduce the occurrence ofunplanned outages. This supports the idea that the immediate short term priority is to improvesystem reliability and establish processes to ensure that all customers are informed of any plannedoutages in advance.

In certain categories of industries, momentary losses form a significant portion of the losses dueto unplanned outages. This supports the idea that the very event of sudden loss of supply is moredetrimental to the operation of an industrial installation than the duration of how long the poweroutage persists. In economic terms, the momentary outage may be viewed as a fixed cost thathappens each time there is an unplanned outage that is independent of the duration of the poweroutage.

Further, the study revealed that a large proportion of industrial facilities have invested instandby generation. In some cases, such as the glass industry, self-generation is used for normaloperation, and power from the utility is used as standby power. This has lead to not only ad-ditional costs in the industrial output but also increased environmental emissions. Therefore, it isclear that the utility needs to take immediate steps to improve its supply reliability in order toretain consumers and justify the existence of a centralised generation facility.

Acknowledgements

The authors are grateful to the USAID SARI/Energy programme and its technical contractor,Nexant, for extending financial assistance for the above study. Also, they gratefully acknowledgethe technical and other forms of assistance extended by the Centre for Energy Studies, Universityof Moratuwa, and the Ceylon Electricity Board.

References

[1] Long Term Generation Expansion Plan 2002–2016. Generation Planning Division, Ceylon Electricity Board, 2001.

[2] Wijayatunga PDC, Attalage RA. Sri Lanka electricity industry: long term thermal generation fuel options.

Institute of Policy Studies, Sri Lanka, 2002.

[3] Annual Report 2000. Central Bank of Sri Lanka, April 2001.

[4] Meir P, Munasinghe M. Incorporating environmental concerns into power sector decision-making: a case study of

Sri Lanka. The World Bank, 1992.

[5] Strait PT. Probability and statistics with applications. New York: Harcourt Brace Jovanovich, Inc; 1983.

[6] Cost of unserved energy. Report No 98PG42, Tata Energy Research Institute, New Delhi, 2001.

[7] Cost of electricity outages in Bangladesh (July 1997–June 1998) by Bangladesh Centre for Advanced Studies

(BCAS) and Unnayan Shamannay (US), Research Report, World Bank, July 2000.

[8] Wacker G, Billinton R. Customer cost of electric service interruptions. Proc IEEE 1989;77(6):919–30.

[9] Kariuki KK, Allan RN. Assessment of customer outage costs due to electric service interruptions: residential

sector. IEE Proc Gener Trans Distrib 1996;143(2):163–70.


[10] Kariuki KK, Allan RN. Evaluation of reliability worth and value of lost load. IEE Proc Gener Trans Distrib

1996;143(2):171–80.

[11] Kariuki KK, Allan RN. Application of customer outage costs in system planning, design and operation. IEE Proc

Gener Trans Distrib 1996;143(4):305–12.

[12] Kariuki KK, Allan RN. Factors affecting customer outage costs due to electric service interruptions. IEE Proc

Gener Trans Distrib 1996;143(6):521–8.

Documents

Assessment of economic impact of electricity supply interruptions in the Sri Lanka industrial sector