Transport Policy 47 (2016) 178–188

Contents lists available at ScienceDirect

Transport Policy

http://d0967-07

n CorrE-m

journal homepage: www.elsevier.com/locate/tranpol

Operational and financial performance of Delhi's natural gas-fueledpublic bus transit fleet: A critical evaluation

Christian Krelling a, Madhav G. Badami b,n

a School of Urban Planning, McGill University, Macdonald-Harrington Building, 815 Sherbrooke Street West, Montreal, QC, Canada H3A 0C2b School of Urban Planning and McGill School of Environment, McGill University, Macdonald-Harrington Building, 815 Sherbrooke Street West, Montreal, QC,Canada H3A 0C2

a r t i c l e i n f o

Article history:Received 23 June 2015Received in revised form5 February 2016Accepted 5 February 2016Available online 23 February 2016

Keywords:Public transitBus transitDelhiNatural gasTransit performancePolicy analysis

x.doi.org/10.1016/j.tranpol.2016.02.0010X/& 2016 Elsevier Ltd. All rights reserved.

esponding author.ail address: madhav.badami@mcgill.ca (M.G. B

a b s t r a c t

Following a Supreme Court of India directive, the bus fleet of the Delhi Transport Corporation (DTC) wasconverted to run on compressed natural gas (CNG) from around 1999 to 2000, to address the city's airpollution. We critically evaluate the operational and financial performance of DTC's bus fleet from 1989–90 to 2010–11 – that is, from ten years prior to CNG implementation until 10 years after – to assess howthis performance was affected by the fuel switch, as well as the introduction of low-floor CNG buses.

CNG implementation caused a significant reduction in the capacity to deliver transit service at DTC inthe initial stages of the fuel transition. Also, it necessitated investments in buses at a considerable costpremium relative to their diesel counterparts. Operating costs per kilometre grew, due to increased fuelexpenditures per kilometre, because of the lower fuel economy, and increased maintenance costs andbreakdowns per kilometre, on the CNG buses. These costs were further exacerbated by the introductionof the low-floor CNG buses. Despite increased capacity due to the investments in the CNG buses, pas-senger-kilometres generally declined over our analysis period. As a result, operating costs per passenger-kilometre, and the ratio of operating costs to traffic revenues, have progressively worsened.

We conclude that the financial situation resulting from these effects due to CNG implementation mayhave detracted from the ability to enhance transit capacity and provide transit service overall. Our studyalso demonstrates the need to analyze policies such as CNG implementation broadly, in terms of conflictsand trade-offs between environmental, and other (transit operation, socio-economic and equity) ob-jectives, rather than narrowly in terms of only environmental outcomes.

& 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Indian cities have been characterized by poor air quality sincethe 1990s. In Delhi, for example, suspended particulate matterlevels have exceeded World Health Organization (WHO) guidelinelimits almost daily since the 1990s. Levels of PM10 (particulatesbelow 10 μm diameter), which are strongly linked with respiratoryand cardio-vascular illnesses and deaths, also exceed the WHOlimits (CPCB, 2015). A global survey of urban air pollution (WHO,2014) showed that Delhi had the highest annual average levels offine particulates (PM2.5), which pose the most serious health risk.In response to this problem, a wide range of policies has beenimplemented since the early 1990s to address air emissions fromurban transport. Delhi being the national capital, and given itsserious air quality problems, many of these policies were firstimplemented there and in the other major metropolitan centres,

adami).

and then in the rest of the country in a phased manner. Thesepolicies have included increasingly stringent vehicle emission andfuel quality standards, vehicle inspection and maintenance (I&M)to control in-use emissions, and the phasing out of old commercialvehicles (CSE, 2002; BIS, 2002; TERI – Tata Energy Research In-stitute, 2002; Kojima, Brandon and Shah, 2000). A Supreme Courtof India ruling in 1998 mandated that all public and for-hire motorvehicles (buses, taxis and auto rickshaws) in Delhi be powered bycompressed natural gas (CNG) (Supreme Court of India, 1998).

As a consequence of this ruling1, all of the city's buses, in-cluding those that were publicly owned and operated by the DelhiTransport Corporation (DTC), had to be converted to run on CNGover a highly compressed time frame, by March 31, 2001. Due toresource, logistical, and institutional challenges, discussed later,

1 According to the ruling (Supreme Court of India, 1998), no 8-year old busescould ply in Delhi except on CNG (or “other clean fuels”) beyond April 1, 2000, andfurther, the entire bus fleet in Delhi was required to be “steadily converted” to runon CNG by March 31, 2001.

(a) Fleet size and age (b) Carrying Capacity-kilometres & Passenger-kilometres

0

1

2

3

4

5

6

7

8

9

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

1989

-90

1991

-92

1993

-94

1995

-96

1997

-98

1999

-00

2001

-02

2003

-04

2005

-06

2007

-08

2009

-10

Ave

rage

flee

t age

(yea

rs)

Num

ber o

f bus

es

Fleet size Buses added Buses scrapped Avegage fleet age

0

20

40

60

80

100

120

0

5,000

10,000

15,000

20,000

25,000

1989

-90

1991

-92

1993

-94

1995

-96

1997

-98

1999

-00

2001

-02

2003

-04

2005

-06

2007

-08

2009

-10

Per

cent

age

Mill

ion

kilo

met

res

Carrying capacity-kilometres Passenger-kilometres Load factor %

Fig. 1. Service provision and utilization.

C. Krelling, M.G. Badami / Transport Policy 47 (2016) 178–188 179

CNG implementation on Delhi's buses began only in 1999–2000. Inany event, Delhi today has the largest bus fleet in India, currentlynumbering, according to official statistics, around 60,000 buses(GNCTD – Govt. of NCT of Delhi, 2012) – and given that all of thecity's buses run on CNG – one of the largest bus fleets running onthis, or indeed any alternative fuel, globally. Further, DTC, the focusof our study, has the second largest publicly owned and operatedurban bus fleet in India, with around 5800 CNG buses currently inoperation (GNCTD – Govt. of NCT of Delhi, 2012). It serves theNational Capital Territory of Delhi, as well as neighbouring cities insurrounding states, and carried 4.5 million passengers daily in2010–11 (CIRT, 2012).

The literature on the implementation of CNG in Delhi's public ve-hicles has focused almost exclusively on its emissions outcomes (forexample, Kathuria, 2005; Jalihal and Reddy 2006; Chelani and Devotta,2007; Reynolds and Kandlikar, 2008; Reynolds et al., 2011; Narain andKrupnick, 2007; Kumar and Foster, 2009), with little if any attentiondevoted to its operational or financial aspects. Sen et al. (2007) discussthis gap in research, pointing to the lack of sufficient focus on the partof decision-makers on the creation of financially viable and self-sup-porting urban transport systems. Further, we argue, while the emis-sions outcomes of CNG implementation are important from a societalperspective, and are an important focus for policy evaluation, it is alsoimportant and useful to critically evaluate this, and indeed, any suchpolicy from the perspective of vehicle users and operators, because itis the policy responses of these actors that crucially determine theextent to which implementation, and the associated emissions re-ductions, actually occur and are successful.

More particularly, analyzing the operational and financial per-formance of bus fleets on CNG or any other alternative fuel isimportant because it is this performance that critically determinesthe bus operator's policy responses. Such an analysis is also usefulfrom a policy perspective, given the need to provide quality,convenient and affordable bus services within the constraints oflimited budgets on which public transit operators typically rely, toprevent the migration of ridership to private motor vehicles, withall of their negative impacts. In this regard, note that, while busesand other public transit modes still account for a significant shareof passenger trips (43% in Delhi), their mode share has been de-clining significantly, due to the growing role of personal motorvehicles (WSA – Wilbur Smith Associates, 2008).

In view of the foregoing, we critically evaluate the operationaland financial performance of DTC's CNG-fueled bus fleet, in orderto assess how the conversion from diesel to CNG has affected fleetoperations and finances. We also critically discuss the im-plementation experience in terms of the associated infrastructure,

logistical and institutional challenges. The 10-plus years of ex-perience accumulated by DTC, a major public transit operator, ofsuch a large-scale conversion of its bus fleet to CNG, provides avaluable opportunity for this retrospective analysis. Apart fromaddressing an important research need (such a post-im-plementation assessment has not been reported on so far), ouranalysis will hopefully be useful to decision makers and urban bustransit operators in contexts similar to India's, by drawing lessonsfor the long-term viability of large-scale conversions of bus fleetsto CNG, for comparison with other CNG bus transit fleets, and forinforming techno-economic and environmental analyses of CNGbus transit operations. In particular, our study, coupled with othersthat have focused on its emissions outcomes, should help assessthe cost-effectiveness of Delhi's CNG policy.

1.1. CNG implementation on DTC's bus fleet

Despite the 1998 Supreme Court mandate to convert the entirebus fleet in Delhi to run on CNG by March 31, 2001, only around 150CNG buses had been put into service at DTC by then. Implementationgained momentum only in 2001–02, and was completed only by2003–04 (Fig. 1A). There were serious logistical and technologicalchallenges related to the large-scale fuel-system conversion that DTC(and even more so, Delhi's private bus operators) faced, particularlywith retrofitting diesel engines to run on CNG. Therefore, they mostlyopted for factory-built CNG buses, which were supplied by two of thelargest Indian bus manufacturers, Tata Motors and Ashok Leyland.More generally, there was a “sequencing problem”, namely that ofimplementation depending on the availability of financial resourcesfor conversion as well as reliable refueling infrastructure and vehicletechnology (Bell et al., 2004). The bus manufacturers and infra-structure providers wanted assurances of demand for the technology,without which they were reluctant to invest in production. In orderto break this supply-demand vicious-cycle, DTC, the bus manu-facturers and Indraprastha Gas Limited (IGL), who provided thededicated refueling infrastructure in DTC's depots, created a taskforce to coordinate their respective roles. Despite the challenges, CNGimplementation was accomplished on Delhi's bus fleet over a shortperiod, from 2001 to 2004 (Bell et al., 2004; Patankar and Patward-han, 2006).

Table 1 highlights the key vehicle and engine attributes of thediesel and CNG bus technologies used at DTC during our analysisperiod. The first generation of CNG buses, which we refer to as theStandard CNG model, were introduced from 2000 to 2004, whenthey replaced a fleet of similar configuration diesel buses (Table 1),with most of these CNG buses being inducted in the 24-month

Table 1CNG bus Technologies at DTC.Source: Delhi Transport Corporation.

Diesel standard CNG standard CNG LF CNG LF/AC

Model year 1997/98 2001/02 2007/08 2007/08Fuel type Diesel CNG CNG CNGEngine model max HP Hino, Leyland 370, Tata 692

95-110IVECO 8060.05; TATA LPO 1510 CGS/55 120-150

Cummins LB BGe230/30 230 Cummins LB 250

Compression ratio 17:1 – 17.9:1 10.5:1 10.5:1 10.5:1No. of cylinders 6 6 6 6Engine displacement 5.9 L 5.9 L 5.9 L 5.9 LEmissions control/standards 1996 norms (pre BS-I) BS-I / BS-II catalytic converter BSIII; Engelhard Oxidation CatalystFuel capacity 160–165 L 90–96 kg 108 kg 108 kgGross Vehicle Weight (GVW) 15.2 15.3 16.2 16.2Length (metres) �10.1 �10.7 12 12Width (metres) 2.4 2.4 2.6 2.6Passenger capacity �65 68 70 70Inflation adjusted bus cost(USDx103)

47.4 59.6 126 149

% of fleet on road (2010–11) 1% 46% 43% 10%

C. Krelling, M.G. Badami / Transport Policy 47 (2016) 178–188180

period from 2001–02 to 2002–03. From the transit user's per-spective, the quality of the buses changed little due to this in-troduction, apart from tailpipe emissions due to the substitution ofdiesel fuel with natural gas. From 2007–08, DTC began a fleetmodernization process by launching two procurement cycles forLow Floor (LF) CNG buses. The introduction of LF CNG buses, whichcommenced towards the end of 2007, was part of a process thatsought to replace the fleet of ageing standard CNG buses, 10% ofwhich were over DTC's target service age2 in 2006–07 (CIRT,2008), while also offering higher quality bus service generally, andin particular, improved and more convenient accessibility for theaged and infirm, as well as young children. Key bus attributechanges included a 400 mm floor height for LF versus 800 mm forstandard buses, automatic transmission, full air suspension, airconditioning (AC) option for higher fares, and a target service lifeof 12 years (or 750,000 km). From 2007 to 2011, 3,700 new LFbuses were put into operation at DTC, of which 67% were non-airconditioned buses (CNG LF) and 33% were air-conditioned (CNGLF/AC) (Table 1). In 2010–11, of the 4300 operational buses, nearly46% were standard CNG buses, 53% were LF buses, and less than 1%of the buses were diesel powered, which were in the process ofbeing decommissioned.

The standard CNG buses, introduced in 2000, had a spark-ig-nited engine with a power rating 31% higher than its standarddiesel counterpart, and CNG storage in 8–12 gas tank cylinderslocated under the passenger floor. Their purchase cost was 26%higher than that of the diesel buses they replaced. The LF CNGbuses included an integral body chassis, rather than being built ona truck chassis as was the case for the standard CNG and dieselbuses, a larger floor area, rear mounted engines with nearly doublethe power rating, and roof-mounted CNG storage tanks. Also, thesebuses conformed to the Bharat Stage III and IV emission standards,which are comparable to EURO-III and IV norms. The LF CNG busescost USD 126,000–149,000, more than twice the average price ofUSD 60,000 for a new standard CNG bus, based on recent

2 The target service life of standard CNG bus set by DTC was 8 years or 500,000km (CIRT, 2012).

procurements by other Indian transit operators (Government ofIndia, 2009a, 2009b, 2009c, 2011, 2012).

Operation, storage and maintenance of DTC's bus fleet areconducted from 49 depots across Delhi. The majority of thesedepots have dedicated CNG refueling infrastructure on site, whichincludes diesel-powered compressors, CNG buffer storage tanks,and usually two fast-fill dispensers per depot. Natural gas feed-stock costs, staffing, operation, and maintenance of DTC's CNGrefueling infrastructure is provided by IGL, which sells CNG to DTCat a negotiated price. Each depot serves only one type of bustechnology (i.e., standard CNG, or LF CNG buses). DTC staff main-tain the standard CNG buses at the depots, while overhauling andreconditioning is done at two central workshops.

When the LF CNG buses were procured, starting in 2007–08,DTC contracted with the bus manufacturers, requiring them to beresponsible for maintaining these buses. This outsourcing was aclear departure from the traditional practice of “in-house” main-tenance, as was the case when CNG was first implemented in2000, and allowed for technological risk to be transferred to thebus manufacturers while allowing DTC to reduce staffing needssubstantially and focus staffing resources on transit operations.Under the agreement with DTC, the bus manufacturers employtheir staff for routine and preventive maintenance of the LF CNGbuses on site at the depots, and bear the associated costs of labourand parts. They carry out overhauls and more specialized main-tenance at outside locations.

2. Analytic framework, methodology and data

In this paper, we present our analysis of DTC's bus fleet per-formance over the period 1989–90 to 2010–11 – that is, fromroughly ten years prior to CNG implementation until 10 years after– in order to assess how fleet operational and financial perfor-mance was affected by the replacement of diesel with CNG, andchanges to bus technology. Our analysis was based on those factorsand performance measures most affected by fuel system change;namely, bus transit service provision and utilization (in terms offleet size and age, carrying capacity-kilometres, and passenger-

C. Krelling, M.G. Badami / Transport Policy 47 (2016) 178–188 181

kilometres); fuel economy (in energy equivalent terms, for dieselversus the CNG operation, and for standard CNG versus LF CNGversus LF-AC CNG buses); capital and operational expenditures(related to fuel, maintenance and labour); and reliability (in termsof breakdowns, for the fleet, and for standard CNG versus LF CNGbuses).

The data for our analysis was based on annual statistics gath-ered at DTC by the authors during a field trip from 02/2010 to 04/2010, by compiling monthly reports published internally for DTCmanagement, as well as on yearly reports published by the CentralInstitute of Road Transport (CIRT, 1992–1995; CIRT, 1997–2012).Unless stated otherwise, all annual time series data were for therelevant financial year (so, a statistic for period 2001–02 refers todata from April 1, 2001 to March 31, 2002). Further, the data weused and our calculations focused only on DTC's bus operations,excluding those related to buses they hired (shown as DTC Hiredin the CIRT reports).

When calculating operational and financial performance mea-sures on a per-kilometre basis, we used effective or revenueearning kilometres, henceforth referred to as kilometres. Fueleconomy figures were as reported by CIRT and DTC (see below),and were based on gross kilometres. CNG fuel consumption wasconverted to litres diesel equivalent, based on the mass basedenergy content of the two fuels, and the density of diesel in Delhi(SIAM – Society of Indian Automobile Manufacturers, 2013). Allfinancial parameters reported in Indian Rupees (INR) were, unlessotherwise stated, inflation adjusted based on the Consumer PriceIndex for Industrial Workers (CPI-IW) published by the LabourBureau of the Government of India (Government of India, 2013),with 2010–11 as the base year (the average annual inflation rateduring our 20-year analysis period ending 2010–11 was 7.63%).Finally, values expressed in US dollars (USD) were derived fromthe inflation adjusted INR values, using the average exchange ratefor 2010–11 (Reserve Bank of India, 2013).

While the annual DTC and CIRT data were useful for analyzingfinancial and operational performance, they were mostly ag-gregated for the entire fleet. Higher frequency data, available on amonthly basis, were accessed from DTC and used for analyzingfleet fuel economy and reliability, and to compare different bustechnologies that were operational contemporaneously. The re-ports accessed at DTC provided statistics on aggregate fleet per-formance, as well as statistics disaggregated at the depot level forkey operational and financial parameters such as staffing, routes,fleet size, kilometres operated, key expenditure items, and mate-rial consumption (fuel, tyres, oil). Information was also gatheredfrom DTC managers at selected bus depots, relating to systemscharacteristics, maintenance, refueling infrastructure and vehicletechnology, fuel economy (FE) and reliability. The sites at whichdata was collected were chosen to ensure representativeness ofthe geographical regions served by DTC's urban transit routes, andaccounted for the fact that DTC operated two types of CNG bustechnologies in 2010, for which bus attributes varied significantly.3

Finally, interviews were conducted with DTC management, toelicit their perspectives based on their experience with CNGimplementation.

3 The DTC sites visited for data collection were the Strategic Business Unit,DTC’s management office for the LF bus fleet, in Hauz Khas; DTC’s Central Work-shop I on Banda Bahadur Marg, where planning for the standard buses, and re-conditioning, re-treading and bus body repairs, were conducted; the Kalkaji Depotand Hari Nagar Depot II (operating the standard buses), and the Sukhdev ViharDepot, Hari Nagar Depot I and Rohini Depot I (operating the LF buses).

3. Results and discussion

3.1. Operational performance

3.1.1. Service provision and utilizationFig. 1a shows the buses added and scrapped, and the resulting

fleet size and age, while Fig. 1b shows the changes in carryingcapacity-kilometres4 and passenger-kilometres over our 20-yearanalysis period. From 1989–90 to 1999–00, when DTC's buseswere exclusively diesel powered, there was no significant capitalinvestment in fleet renewal, and the average age of DTC's fleetsteadily increased to almost 8 years, DTC's benchmark for max-imum service life for standard buses. From 1999–00 to 2002–03,as highlighted in Fig. 1a, DTC's fleet went through a transitionperiod, in which fleet capacity was affected by, among other fac-tors, the Supreme Court's scrappage and CNG mandates. Thescrappage mandate requiring the retirement of buses older thaneight years preceded the mandate that all buses be converted toCNG by one year (see Footnote 1).

As a result of these mandates, DTC's bus fleet size, which wasalready in decline, was reduced precipitously, by about one-third,from 3000 buses to 2000 buses, during the transition period(Fig. 1a). CNG implementation accelerated from 2001–02, as noted,and as new CNG standard buses were inducted into the fleet, thefleet size returned to pre-transition period levels, the fleet com-position changed from mostly diesel to mostly CNG, and theaverage fleet age decreased significantly. DTC's CNG bus technol-ogy was predominantly homogeneous, with the fleet being com-prised only of standard CNG buses, and vehicle scrappage and fleetrenewal being low, during 2001–2009, therefore allowing the useof operational statistics aggregated at the fleet level over this timeto analyze the performance of standard CNG buses over theirservice life. CNG LF bus introduction gained momentum in 2009,and fleet-wide performance statistics start to be increasingly in-fluenced by this technology from this time onward. In 2010–11,DTC's fleet comprised around 6200 CNG buses, but only around4300 were in use due to about 1800 standard CNG buses beingover their target service life of 8 years (or 500,000 km) anddeemed unfit for operation (CIRT, 2012).5

Carrying capacity-kilometres, as well as passenger-kilometreswere both already in decline since 1989–90 (Fig. 1b), but with thesharp reduction in the bus fleet due to the scrappage and CNGmandates, there was an equally sharp (40%) reduction in thesemeasures in the beginning of the transition period. Further, theaverage load factor6 was extremely high, reaching almost 100% fortwo consecutive years. With the induction of standard CNG busesfrom 2000–01, carrying capacity-kilometres increased steadilyuntil 2005–06, but interestingly, passenger-kilometres did not atthe same rate. Beyond 2005–06, and until 2009, when the CNG LFbuses were introduced, both capacity-kilometres and passenger-kilometres again dropped precipitously, as the standard CNG busesthat were past their target service life were taken out of service,but also possibly because of the significant increase in busbreakdowns during this period, as we discuss in a subsequentsection. These measures began to revive again, after the in-troduction of the new CNG LF buses. Overall, what is striking is thedecline in passenger-kilometres over the 20-year analysis period,and that this measure remains flat, even while the fleet size was

4 Capacity-kilometres are calculated by multiplying the total seating andstandee capacity of all buses by the kilometres operated (CIRT, 2009).

5 While diesel buses were being used in inter-state operations, these busesaccounted for less than 1% of the operational fleet in 2010–11, and so have limitedimpact on fleet-wide statistics.

6 The load factor is the ratio of passenger-kilometres to carrying capacity-kilometres, expressed as a percentage (CIRT, 2009).

(a) Yearly Fuel Economy for Standard diesel and CNG buses (b) Monthly Fuel Economy, Standard versus LF CNG buses

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

1989

-90

1991

-92

1993

-94

1995

-96

1997

-98

1999

-00

2001

-02

2003

-04

2005

-06

2007

-08

2009

-10 Fu

el E

cono

my

(km

/die

sel l

itre

equi

vale

nt)

0

CNG (km/diesel equivalent litre)

Diesel (km/l)

1.0

1.2

1.4

1.6

1.8

2.0

2.2

2.4

Jan-

09

Feb-

09

Mar

-09

Apr

-09

May

-09

Jun-

09

Jul-0

9

Aug

-09

Sep

-09

Oct

-09

Nov

-09

Dec

-09

Jan-

10 Fu

el E

cono

my

(km

/die

sel l

itre

equi

vale

nt)

Standard CNG buses (19 depot sample)

LF CNG (3 depot sample)

LF-AC CNG (1 depot sample)

Fig. 2. Fleet fuel economy.

C. Krelling, M.G. Badami / Transport Policy 47 (2016) 178–188182

being augmented significantly with new CNG buses.7,8 This trendlikely reflects (and is reflected in) the steady decline in publictransit shares reported in WSA – Wilbur Smith Associates (2008),which is likely due to a range of factors, but also possibly thegrowing role of the Delhi metro.

3.1.2. Fleet fuel economyThe average fuel economy of standard CNG buses was 45% lower

than that of DTC's diesel-only fleet, in km/L diesel equivalent terms(Fig. 2a). This significant reduction in fuel economy in equivalentenergy terms is likely due to several factors. First of all, while theOtto cycle, on which the spark-ignited (SI) CNG engines operate, hasa higher thermal efficiency than the diesel cycle at the same com-pression ratio, diesel engines use much higher compression ratiosin practice, because they compress only air, and are thus not sus-ceptible to auto-ignition (as in the case of SI engines), and aretherefore able to achieve higher thermal efficiencies than SI en-gines. CNG is a much better SI engine fuel than gasoline, because ofits very high octane rating, and therefore SI CNG engines can use acompression ratio higher than that of gasoline engines, but still,much lower than that of diesels (in the case of DTC's buses, as canbe seen from Tables 1, 10.5 for the SI CNG engines, relative to the17.1–17.9 for the diesels which they replaced). Additionally, SI en-gines are characterized by poor part-load efficiencies because ofthrottling losses. In the case of the DTC CNG buses, it is also possiblethat sub-optimal engine technology, as well as the additionalweight of the CNG fuel tanks contributed to the significantly lowerfuel economy. Yet another important factor is likely the low speed,stop–go characteristics of DTC bus operations; note in this regardthat congestion is a major and growing challenge in Delhi, wherethe average peak-hour speed – for all modes – was 16 km/h, ac-cording to WSA – Wilbur Smith Associates (2008). Chassis dy-namometer and road tests have shown that the fuel economy ofCNG buses with similar specifications to diesel buses are typically4–25% lower depending on the vehicle operating cycle (Lowell et al.2007; Wayne et al. 2008; Clark et al. 2009), with the gap beinggreater at slower average speeds.9

7 One factor contributing to the low passenger-kilometres from 2000 onward isthat the average annual distance covered by DTC’s CNG buses was approximately74,000 km (CIRT 2012), as opposed to 82,338 km on its diesel buses during the1990s, according to CIRT data. Also, note that the corresponding figure in otherIndian metro cities is 85,000 km (as against only 56,000 km in US cities – Lowellet al. 2007).

8 Note that, while the average load factor for DTC was 71% from 2000–01 to2010–11, it was 64% for Bangalore’s and Mumbai’s public bus fleets, and 81% forChennai’s (CIRT, 1992–1995; CIRT, 1997–2012).

9 The US experience has shown that, because SI CNG engines are far more

While the fuel economy of DTC's fleet reduced significantly as aresult of the switch to CNG on their standard buses, it deterioratedeven further with the introduction of LF buses in 2007, a processthat gained momentum in 2009, as noted. The key differences inthe LF CNG buses relative to the standard CNG buses included aone tonne higher vehicle weight, and a 70% higher engine powerrating (Table 1). Based on data collected at DTC, the average fueleconomy was 17% and 35% lower for the LF (non AC) and LF-ACCNG buses respectively, than for the standard CNG buses duringthe same evaluation period in 2009–10. In the case of the AC bu-ses, there was greater seasonal variation due to the air con-ditioning system.

3.1.3. ReliabilityOne measure of service reliability is the rate of breakdowns.10

In order to evaluate the reliability of alternative fuel systems onbuses, it would be useful to focus on breakdowns that are relatedto the fuel system, engine, air intake, cooling, and exhaust. How-ever, the CIRT breakdown statistics (CIRT, 2001–2012) are dis-aggregated only in terms of mechanical, electrical and tyre relatedbreakdowns, with mechanical breakdowns comprising those re-lated to transmission, engine and brakes. So, for the evaluation ofreliability (in terms of breakdowns) of CNG buses relative to theirdiesel predecessors over our analysis period (Fig. 3a), we used CIRTstatistics, excluding tyre related breakdowns. Since vehicle re-tirement influenced bus availability, and likely distorted the sta-tistics during the transition from diesel to CNG, the period from1999–00 to 2001–02 was excluded from this analysis.

Interestingly, the diesel fleet showed an improving trend inbreakdown rates (Fig. 3a), despite the increasing age profile of thefleet (Fig. 1a), during 1990–91 to 1998–99. On the other hand, thebreakdown rate worsened significantly for the CNG fleet, as thefleet aged, with the average from 2002–03 to 2010–11 being 125%higher than for the diesel fleet. Since effective kilometres per buson road is a stronger determinant of wear and tear, and thereforebreakdowns (and maintenance costs), we have plotted this

(footnote continued)sensitive to operating conditions than their diesel counterparts, higher averagespeeds significantly improve performance for both diesel and CNG buses, but muchmore for CNG buses. Further, the fuel economy of state-of-the-art SI CNG enginesactually approach that of diesels, under favourable operating conditions, and par-ticularly when stringent emission standards call for more advanced exhausttreatment on diesels, with negative implications for their fuel economy – see forexample Melendez et al. (2005) in this regard.

10 A breakdown is defined as stoppage of a bus (while in operation) due tomechanical defects or other failures rendering the bus unfit to operate withoutattention to it, irrespective of the time involved (CIRT, 2009).

Fig. 3. Reliability.

C. Krelling, M.G. Badami / Transport Policy 47 (2016) 178–188 183

measure for reference in Fig. 3a; note that the significantly higherbreakdown rates for the CNG buses occurred despite lower effec-tive kilometres per bus for CNG relative to diesel buses, and asignificant decline in this parameter from 2005–06. Also worthnoting is the significant increase in electrical breakdowns, likelydue to the spark plugs and related electrical systems on the SI CNGengines, a problem that has also been observed elsewhere(Chandler, 2006). New York City Transit (NYCT), which operates afleet of over 4000 buses, including a large number of CNG buses,has set performance criteria, which among other things requirethat buses should be operated at least 6400 km between roadcallson average (Barnitt and Chandler, 2006). This benchmark, which isequivalent to 1.56 breakdowns per 1o,ooo km (for allbreakdowns)11, was exceeded beyond around 2010, when many ofthe standard CNG buses were past their 8-year target service life,and the LF CNG buses were being rapidly inducted into DTC's fleet(see below).

In Fig. 3b, we compare the breakdown rates of the standard andLF CNG buses during the time period December 2007 to December2009, based on monthly breakdown statistics accessed from DTC,which did not discriminate between types of breakdowns, and soinclude all breakdowns. The breakdown rate for the LF buses,which on average was 2.4 breakdowns per 10,000 km, was 41%worse than for the standard CNG buses (and consistently worsethan NYCT's benchmark of 1.56 breakdown-per-10,000 kmthroughout), despite the very low (1.5 years) average age of theformer buses, and roughly 67% of the standard CNG buses beingolder than 8 years during this period. This poor performance onthe part of the LF buses, which possibly is due to the fact that theywere being implemented for the very first time in India when theywere introduced at DTC, has raised public concern and negativepublicity, as evidenced by media reports (for example, The Timesof India, 2013).12 Our discussions at DTC revealed that the busmanufacturers had likely underestimated, and were overwhelmedby the maintenance costs of the LF CNG buses. As for the drop inbreakdown rates for 2010–11 (the last data point in Fig. 3a), notethat the data for 2011–12 (not shown in the figure) in fact goesback up to the 2009–10 level. While Fig. 3a and b, taken together,show higher breakdown rates for the LF CNG relative to the

11 NYCT’s CNG buses had an average breakdown rate, for all types of failures, ofjust 1.1 breakdowns per 10,000 km (Barnitt, 2008).

12 Our research at DTC, based on a sample of 300 LF CNG buses, showed that,from February 2009 to January 2010, mechanical, tyre and electrical problems ac-counted for 45, 27 and 17% of the breakdowns, respectively.

standard CNG buses, it might perhaps be best to wait until around2016–17 in order to properly evaluate LF buses in this regard, overtheir full service life.

3.2. Financial performance

3.2.1. Capital expendituresDTC's capital expenditures during our analysis period from

1989–90 to 2010–11 were mostly affected by the purchase ofroughly 3000 standard CNG buses in 2000–2004, and then by thepurchase of 3700 LF CNG buses starting in 2007–08. There waslimited impact on DTC finances due to other capital expenditures,such as those tied to the refueling infrastructure or changes todepots and maintenance areas for CNG operation. The refuelinginfrastructure was put in by IGL, the gas supplier, as discussedearlier, and so DTC did not incur any directly related capital ex-penditures. However, since DTC pays IGL, which is a publicly tra-ded company, for the fuel that is provided on site, the CNG pricelikely factors in IGL's recovery of their capital investment, and so,DTC does pay indirectly for the refueling infrastructure. While wedid not have access to information on expenditures on depotmodifications related to the CNG refueling infrastructure, weevaluated them indirectly based on the changes to asset values ofbuilding infrastructure at DTC during the transition from diesel toCNG. Despite the value of building, plant and machinery havingincreased by 43% from 2001 to 2004 (based on CIRT data), theseassets accounted for only 12%, whereas DTC's bus fleet did so for asmuch as 85%, of DTC's total fixed assets in 2004. This situation ispartly due to the fact that Delhi's climate allows for open storageof CNG buses and semi-open maintenance areas, which requireonly minimal investment in depot infrastructure. In other con-texts, especially in colder climates, most depot upgrades related toCNG buses are due to installation of ventilation systems, re-wiringof electrical systems, and gas detection systems to mitigate therisks of gas leakages, explosions and fires (Barnitt and Chandler,2006; Chandler et al., 2006).

In terms of bus capital expenditures, purchase costs (Table 1)were based on information collected at DTC, as previously noted.Standard diesel bus costs were based on the last procurement ofthis bus type at DTC in 1998–99, while the costs for standard CNGbuses are the average price from 2000 to 2003, when DTC's fleetwas being converted to this fuel. The cost premium for the stan-dard CNG bus relative to its diesel predecessor was 26%, mostlyrelated to the fuel system and engine, apart from which both bustypes had similar vehicle attributes (Table 1). While DTC has not

(a) Operating expenses, various categories (b) Operating expenses per kilometre

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

1989

-90

1991

-92

1993

-94

1995

-96

1997

-98

1999

-00

2001

-02

2003

-04

2005

-06

2007

-08

2009

-10

Ann

ual E

xpen

ses

(mill

ion

INR

, inf

latio

n ad

just

ed)

Fuel expenses

Other materials (lubricants, springs, tyres, batteries, general & reconditioned items) Auto spare parts

Labour (workshop & maintenance)

Labour (administrative & bus operations)

4

5

6

7

8

9

10

11

12

13

14

1989

-90

1991

-92

1993

-94

1995

-96

1997

-98

1999

-00

2001

-02

2003

-04

2005

-06

2007

-08

2009

-10

INR

per

kilo

met

re, I

nfla

tion

Adj

uste

d

Labour (workshop & maintenance) and materials (lubricants, spares, tyres, batteries, etc.)

Fuel expenses

Fig. 4. Key operating expenses.

13 Data for various transit performance parameters needs to be provided moreconsistently for different parts of DTC (DTC urban, rural, hired, NCT), both to allowthe different parameters to be correlated more effectively, as well as to enable aproper comparison across different transit operators. With specific reference tobreakdowns and maintenance costs, data needs to be provided in a much moredisaggregated fashion, to enable assessment of alternate fuel systems in terms ofthese parameters.

C. Krelling, M.G. Badami / Transport Policy 47 (2016) 178–188184

procured standard buses in recent years, their attributes havechanged little, apart from vehicle emission standards, partial use ofair suspension, and possibly GPS units. The average current cost ofa standard diesel bus in India is USD 60,000, and the cost premiumof a standard CNG bus is 20%, based on data for Indian transitoperators that procured both standard diesel and CNG buses(Government of India, 2009a, 2009b, 2009c, 2011, 2012). The CNGbus premium has therefore not changed significantly since theearly 2000s.

DTC's costs for the non-AC LF CNG buses increased 27%, fromINR 4.1 million (USD 123,000) to INR 5.2 million (USD 126,000) perbus, between the first and second procurements in 2007 and2009; note that inflation was only 8.7% (based on the consumerprice index referred to earlier) over the same period. Meanwhile,costs for the LF AC CNG buses actually decreased 9% over the sameperiod to INR 6.2 million (USD 149,000), probably owing to theincreased scale of DTC's purchase of 775 buses in the secondprocurement, relative to the first one, in which only 25 LF AC CNGbuses were purchased. Assessing the marginal cost of CNG LFbuses relative to their diesel counterparts at DTC is a challengebecause no new diesel buses have been purchased since adoptingCNG. Data from Municipal Corporation Ludhiana (2013), whichprocured similar specification LF non-AC diesel buses at the cost ofINR 4.8 million (USD 105,000) per bus from the same manu-facturer that supplied most buses at DTC, shows that the costpremium of a non-AC LF CNG bus at DTC is about 20%; note,however, that, apart from basic engine and model specifications inthe Ludhiana case, not much is known about other vehicle attri-butes, which also could affect costs. Data for Indian transit op-erators that purchased both CNG and diesel buses (Government ofIndia, 2011) shows that the average cost premium for CNG buseswas 8–25%, with lower premiums for the more expensive semi-LFor LF CNG buses, and higher premiums for the less expensivestandard CNG buses, relative to their diesel counterparts, thuscorroborating the comparison between DTC and Ludhiana. Thesefindings in the Indian context are similar to the cost premiums of8–20% for CNG buses observed in the USA and Europe (Posada,2009).

3.2.2. Operating expendituresThe area graph in Fig. 4a shows the key operating expenditures

from 1989–90 to 2010–11. Counting from the bottom of the figure,the first two wedges, put together, represent the total labour costs;

the second and third wedges represent the workshop and main-tenance labour plus the expenditure on automobile spare parts,and the third and fourth wedges the total expenditures on allmaterials, except fuel, which are shown in the top wedge. Taken asa whole, the expenditures in Fig. 4a represent the total operatingexpenditures, less taxes, interest, depreciation and other mis-cellaneous expenses. In Fig. 4b, we show the fuel costs, and theexpenditures on workshop and maintenance labour plus all ma-terials except fuel, per effective (or revenue generating) kilometre(note that the CIRT data does not indicate the expenditures ex-clusively related to bus maintenance, except for the workshop andmaintenance labour).13 For our analysis, we also compare theprices of diesel and CNG on an energy equivalent basis over theanalysis period, in Fig. 5a.

The first point to note is that interest payments (not shown inFig. 4a) alone amounted to around 50% of total operating ex-penditures. This high debt burden is mostly due to the cost asso-ciated with the purchase of standard CNG buses from 2000–01 to2003–04, and the LF CNG buses starting in 2007–08. The CIRT datashows that these purchases were financed mainly through gov-ernment loans, at annual interest rates of 10.5–14.5%, and withmaturities around 13 years. Given the large CNG bus procurements(which involved significant cost premiums relative to their dieselcounterparts), the high interest rates, and short maturity periods,DTC's liabilities have grown exponentially since CNG was in-troduced. The second key point is that labour expenses, of whichthose related to administration and bus operations form the lion'sshare – and have been increasing continuously, and particularlyrapidly since around 2007–08 – account for the bulk of operatingexpenditures less taxes, interest, depreciation and other mis-cellaneous expenses. An important factor in this regard is labourproductivity. In 2010–11, the average DTC depot, with 830 per-sonnel, serviced 110–130 buses, for an average of 7.5 personnel-per-bus, as against an average of 6.2 personnel-per-bus in theother large public bus transit operations in Bangalore, Chennai,

(A) Fuel prices paid by DTC, annual average (B) CNG-diesel price ratio (PCNG ÷ Pdiesel)

10

15

20

25

30

35

40

45

50

55

1998

-99

1999

-00

2000

-01

2001

-02

2002

-03

2003

-04

2004

-05

2005

-06

2006

-07

2007

-08

2008

-09

2009

-10

2010

-11

Jun-

Jul.'

12

INR

per

litre

(Inf

latio

n ad

just

ed) Diesel price per litre

CNG price per diesel equivalent litre

0.2

0.4

0.6

0.8

1.0

1.2

2001

-02

2002

-03

2003

-04

2004

- 05

2005

-06

2006

-07

2007

-08

2008

-09

2009

-10

2010

-11

Jun-

Jul.'

12

CN

G-d

iese

l pric

e ra

tio(o

n an

ene

rgy

equi

vale

nt b

asis

)

USA

India

0

Fig. 5. Comparative fuel prices – diesel and CNG. Source: India from 1998–99 to 2009–10 (DTC); India from 2009–10 to 2010–11 (CIRT 2011, 2012); India June–July 2012(MyPetrolPrice.com, 2013); USA: EERE (2013).

14 After the fleet had been fully converted to CNG, in 2003–04, standard CNGbuses made up over 90% of the operational fleet, with only a marginal presence ofstandard diesel buses that operated on inter-state routes.

15 Over 1843 standard CNG buses were older that the targeted service life of8 years in 2011 (CIRT, 2012), and most likely contributed to the high maintenancecosts per kilometre (Fig. 4), as well as the higher breakdown rates in Fig. 3a.

C. Krelling, M.G. Badami / Transport Policy 47 (2016) 178–188 185

and Mumbai, and 2.5 personnel-per-bus in the USA (CIRT, 2012;Lowell et al., 2007, Clark et al., 2009).

Thirdly, note that total fuel expenses, being a function of fleetfuel economy, fuel price and total fleet activity, are largely drivenby the variations in the last factor over the analysis period. On theother hand, fuel costs per kilometre, which are a function of fleetfuel economy and fuel price, appear to have been driven largely bythe increase in diesel price for the all-diesel fleet until CNG in-troduction in 2000–01 (Fig. 4a), given that fleet fuel economyvaried only slightly, from 3.72 to 3.8 km/L between 1989–90 and2000–01 (Fig. 3a). Note that, even though capacity-kilometres andpassenger-kilometres were dropping significantly, load factorswere as high as 87% on average, during this period.

Fuel costs per kilometre started to increase rapidly with CNGintroduction; indeed, they were 70% higher for the CNG fleet from2003–04 to 2010–11, that is, after the transition to CNG had beencompleted, than for the all-diesel fleet from 1989–90 to 1998–99.This was despite the high load factors and the increase in dieselprices up to the transition period, on the one hand, and the sig-nificant reduction – by as much as a third – in CNG prices, from thetime of introduction of this fuel, in 2000–01, until around 2008–09(Fig. 5b), besides a much lower (69%) average load factor for theCNG buses than for the all-diesel fleet (Fig. 1b). A key contributoryfactor was of course the much poorer fuel economy of CNG busesrelative to diesel buses – a situation that was exacerbated by theincrease in CNG price from around 2009, just as the LF CNG buses,with even poorer fuel economy than the standard CNG buses, werebeing introduced.

A note on CNG prices: in the US context, natural gas feedstock,as represented by the average city-gate price (EIA – U.S. EnergyInformation Administration, 2013), represented only about aquarter of the final retail price of CNG sold at refueling stations(EERE – Energy Efficiency and Renewable Energy, U.S. Dept. ofEnergy, 2013). Other major costs include those related to capitalinfrastructure for refueling stations, energy for compression, fuelretailers' margins, and taxes. All these factors are assumed to bepriced into the CNG price paid by DTC to IGL. The average energyequivalent price of CNG hovered just under 50% of the diesel pricefrom 2001 to 2011 (Fig. 5a). Fig. 5b compares the CNG price re-lative to that of diesel in the US and Indian contexts. It appearsthat, although the CNG-to-diesel price ratio has been higher in theUSA than in India, this ratio has converged to similar levels to-wards the end of 2011.

The expenditures onworkshop and maintenance labour plus all

materials except fuel, per effective kilometre, show large varia-tions over our analysis period (Fig. 4b). Interestingly, the graph forthis measure appears to be a near mirror image of the graph fortotal fleet activity (in terms of capacity-kilometres and passenger-kilometres) in Fig. 1b. This is because workshop and maintenancelabour and material expenses – and particularly workshop andmaintenance labour, which forms the bulk of these expenses – arefairly constant regardless of the fleet activity (or even the fuelsystem), as evidenced by Fig. 4a. Even so, the average maintenanceexpenditure per kilometre for the CNG fleet, from 2003–04 to2010–11, was roughly 16% higher than for the all-diesel fleet, from1989–90 to 1998–99. This is despite the fact that the average ageof the latter fleet was 6.1 years (with a minimum of 5.6 years), andwas almost 8 years before conversion to CNG (Fig. 1a), whereas theaverage fleet age decreased to less than 2 years post-conversion,and gradually increased over the years until the procurement ofthe LF CNG buses.14 Note also the significantly higher breakdownrates for the CNG buses, despite the effective kilometres per bus onroad being lower on average for these buses, and having declinedsignificantly from 2005 to 06, which we discussed earlier (Fig. 3a).A final trend worth noting in Fig. 4b is the decline in the per-kilometre expenditures on workshop and maintenance labour plusall materials except fuel from their peak in 2008–09. This waslikely possible, despite the higher breakdown rates for the LF CNGbuses, mostly due to the fixed nature of the LF bus maintenancecosts under the service contracts, thus highlighting an effectivecost management strategy used by DTC to transfer risk on the newbus technology.

At the same time, though, fleet age likely contributed tomaintenance costs. The period from 2002 to 2009 may be used tounderstand the evolution of these costs for standard CNG buses, asthey aged over their 8-year service life. Analysis of the differentcomponents of these costs (Fig. 6) shows that spare parts ex-penditures, which were initially low, rose to match or even exceedthe labour costs in the last four years of this period.15

As we discussed earlier, passenger-kilometres have declinedover the 20-year analysis period. Whatever the cause of this trend,

0

2

4

6

8

10

12

14

16

0

0.5

1

1.5

2

2.5

3

3.5

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

YR1 YR2 YR3 YR4 YR5 YR6 YR7 YR8 YR9 YR10

INR

per

km

(inf

latio

n ad

just

ed)

INR

per

km

(inf

latio

n ad

just

ed)

Year and Average Standard CNG fleet age

Lubricants Bus spare parts Batteries and other items Labour

Fig. 6. Maintenance costs, key categories, per kilometre.

C. Krelling, M.G. Badami / Transport Policy 47 (2016) 178–188186

the fact remains that it is happening even as CNG was introduced,and new standard, LF and LF–AC CNG buses were inducted intoDTC's fleet from 2000 to 01 onward, involving significant capitalexpenditures (and cost premiums over their diesel counterparts),in the process considerably increasing DTC's debt burden. At thesame time, operating costs have increased considerably (Fig. 4a),even without considering taxes, interest, depreciation and othermiscellaneous expenses, and even as passenger-kilometres haveremained stagnant (Fig. 1b). This effectively means that operatingcosts per passenger-kilometre have been increasing since CNGintroduction; indeed, as shown in Fig. 7, they have grown steadilyover the 20-year analysis period. The net result of this, in turn, isthat the operating ratio, which is the proportion of total trafficrevenue accounted for by operating expenditures, and is thus animportant measure of the operational efficiency of the transitservice, has also been increasing steadily over the analysis period,and exceeded 250% during 2007–09 (again, it must be stressedthat taxes, interest, depreciation and other miscellaneous ex-penses are excluded from Fig. 7).

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

1989

-90

1991

-92

1993

-94

1995

-96

1997

-98

1999

-00

INR

per

pas

seng

er-k

m (I

nfla

tion

adju

sted

)

Total operating costs (less taxes, interest, deprecand miscelaneous others), per pass-km

Operating ratio = (labour+material expenses) /

0

Fig. 7. Operating costs per passenge

4. Conclusions and implications

As we argued in our introduction, the 10-plus years of experi-ence accumulated as a result of the large scale conversion of DTC'sbus fleet to CNG provides a valuable opportunity for a post-im-plementation evaluation of the operational and financial perfor-mance of this fuel system, thereby fulfilling an important researchneed. At the same time, this experience enables lessons to bedrawn for the viability of similar large-scale conversions of urbanbus fleets to CNG, and for informing techno-economic and en-vironmental analyses of CNG bus transit operations, in India andother rapidly motorizing low and middle income countries. So,what lessons may be drawn from the DTC experience?

The fact that the major technical, logistical and institutionalchallenges associated with this implementation were overcome,and CNG was implemented on buses (and other public motorvehicles) in Delhi over the space of 3–4 years, despite thesechallenges, is undoubtedly an achievement, which demonstratesthe importance of co-ordination between various actors, includingvehicle manufacturers, fuel suppliers, bus and other public motorvehicle operators, and different levels of government.

At the same time, it is important to recognize that the im-plementation of the scrappage and CNG mandates, which weredriven primarily by environmental concerns related to Delhi's airquality, caused serious reduction in the capacity to deliver transitservices at DTC (as well as on the part of Delhi's private bus op-erators, and auto-rickshaws) in the initial stages of the transitionto CNG, thereby likely causing significant hardships for commuters(reflected in the very high load factors during 1999–2001), andcompromising easy and affordable access to livelihoods and otheressential services. Ironically, this drastic capacity reduction alsolikely caused personal motor vehicle use to increase during thistime. These unintended effects demonstrate the importance ofanticipating and preparing for them prior to implementation, andalso perhaps, of carrying out the implementation in a phasedmanner.

Above and beyond these effects is the fact that CNG im-plementation necessitated significant investments in buses at aconsiderable cost premium relative to their diesel counterparts,besides the investments in fuel infrastructure and depot

0

50

100

150

200

250

300

2001

-02

2003

-04

2005

-06

2007

-08

2009

-10

perc

enta

ge

iation,

(traffic revenue)

r kilometre and operating ratio.

C. Krelling, M.G. Badami / Transport Policy 47 (2016) 178–188 187

modifications (which were thankfully not as substantial as incontexts such as New York's). Additionally, operating costs perkilometre grew considerably with the introduction of CNG, firstlydue to significantly increased fuel expenditures per kilometre –

despite a low and declining CNG price – because of the lower fueleconomy on the CNG buses, and secondly, increased maintenancecosts (and breakdowns) per kilometre, despite declining effectivekilometres per bus on road. Both of these costs were further ex-acerbated by the introduction of the LF and LF-AC CNG buses.Taken as a whole, these factors have adversely affected DTC's fi-nancial situation.

Further, despite enhancements to capacity-kilometres as a re-sult of the significant investments in standard CNG buses during2000–04, and then in the LF and LF-AC CNG buses – which pre-sumably were introduced to make transit more attractive and in-crease ridership – in 2007–10, passenger-kilometres generallydeclined over our 20-year analysis period. As a result, operatingexpenditures per passenger-kilometre, and the ratio of operatingexpenditures to traffic revenues, both important measures of op-erational efficiency of the transit service, have progressively wor-sened. This situation is only likely to be further exacerbated byCNG prices increasing to international levels.

The question that our analysis raises is, apart from its emissionsoutcomes, how has the significant investment in CNG contributedto, or taken away from, the objective of providing convenient, af-fordable and viable public transit service in Delhi, which is cru-cially important first and foremost to cater for the accessibility andmobility needs of the masses, and also to minimize the need forpersonal motor vehicle activity and its associated impacts, in-cluding air pollutant emissions. This question gains particularimportance because of the decline in passenger-kilometres at DTC,and the equally significant reduction in public transit modal sharesin Delhi among other metropolitan cities in India, over the pastcouple of decades.

It may be argued that the debt burden, and the overall financialsituation due to the increased capital and operating costs perpassenger-kilometre due to CNG implementation has in fact de-tracted from this objective, as well as the ability to enhance publictransit capacity and provide widespread coverage region-wide. Itis worth investigating whether, if the same investment had beenmade in emission-controlled diesel buses, a larger capacity todeliver bus transit services might have resulted, thereby betterachieving equity, and conceivably, even environmental objectives,by helping avoid a larger number of personal motor vehicle tripsregion-wide. An additional problem in this regard is of course thatthe investments are being made without any accompanying TDMmeasures.

All of the foregoing points have important implications forpolicy-making and implementation, as well as for policy analysis.They demonstrate the need to analyse and formulate environ-mental policies such as the implementation of CNG on Delhi'sbuses broadly, in terms of a wide range of impacts for differentgroups in society, rather than narrowly in terms of only environ-mental (in this case emissions) outcomes, and to explicitly con-sider and address conflicts and trade-offs between environmental,and other (transit operation, socio-economic and equity)objectives.

Acknowledgements

We are grateful to the Social Sciences and Humanities ResearchCouncil (Grant no. SSHRC 410-2005-2390), the International De-velopment Research Centre (Award no. 105938-99906075-056),and McGill University for funding the research on which this paper

is based. We thank Professors Geetam Tiwari and Dinesh Mohan ofthe Transportation Research and Injury Prevention Programme(TRIPP), IIT Delhi, and Professor Sanjivi Sundar of The Energy andResource Institute (TERI) and Dr. Rajiv Seth of TERI University, forhosting the first author during his field-work. We owe a debt ofgratitude to Mr. V.K. Sehgal and Mr. S.P. Sethi for providing accessto data, and for discussions, at Delhi Transport Corporation (DTC).Lastly, the second author is grateful to IIM Bangalore, UniversityCollege London, and The University of Manchester, where he wasbased during his sabbatic leave, while the paper was revised andfinalized for submission.

References

Barnitt, R., Chandler, K., 2006. New York City Transit (NYCT) hybrid (125 Order) andCNG transit buses final evaluation results. National Renewable Energy La-boratory, Golden, Colohttp://dx.doi.org/10.2172/894985.

Barnitt, R., 2008. In-Use Performance Comparison of Hybrid Electric, CNG, andDiesel Buses at New York City Transit. SAE International Powertrains, Fuels &Lubricants Conference, June 2008, Shanghai, China. SAE 2008-01�1556.

Bell, R.G., Mathur, K., Narain, U., Simpson, D., 2004. Clearing the air: How Delhibroke the logjam on air quality reforms. Environment 46 (3), 22–39.

BIS, 2002. Motor Gasolines – Specification, IS 2796:2000. Bureau of Indian Stan-dards, New Delhi.

Chandler, K., 2006. Ten years of compressed natural gas (CNG) operations at Sun-Line Transit Agency, April 2003–December 2004. National Renewable EnergyLaboratory, Golden, Colohttp://dx.doi.org/10.2172/876662.

Chandler, K., Eberts, E., Melendez, M., 2006. Washington Metropolitan Area TransitAuthority compressed natural gas transit bus evaluation. National RenewableEnergy Laboratory, Golden, Colohttp://dx.doi.org/10.2172/881923.

Chelani, A., Devotta, S., 2007. Air quality assessment in Delhi: before and after CNGas fuel. Environ. Monit. Assess. 125 (1), 257–263. http://dx.doi.org/10.1007/s10661�006�9517-x.

CIRT, 1992 to 1995. Performance Statistics of State Transport Undertakings. Reportsfor the years 1989-90 & 1990-91, 1990-91 & 1991-92, and 1992-93 & 1993-94.Central Institute of Road Transport, Pune, India.

CIRT, 1997 to 2012. State Transport Undertakings - Profile and Performance. Reportsfor the years 1994-95 & 1995-96, 1995-96 & 1996-97, 1997-98, 1998-99, 1999-2000, 2000-01, 2001-02, 2002-03, 2003-04, 2004-05, 2005-06, 2006-07, 2007-08, 2008-09, 2009-10, 2010-11. Central Institute of Road Transport, Pune, India.

CIRT, 2009. Compendium of Transport Terms, Seventh Revised Edition. CentralInstitute of Road Transport, Pune, India.

Clark, N., Zhen, F., Wayne, W., Schiavone, J., Chambers, C., Goloub, A., Chandler, K.,2009. TCRP Report 132: Assessment of Hybrid-Electric Transit Bus Technology.Transportation Research Board of the National Academies, Washington, D.C.

CPCB (2015). Ambient air quality data. New Delhi: Central Pollution Control Board.Accessible at: ⟨http://cpcb.nic.in/⟩.

CSE, 2002. State of India’s Environment – The Citizens’ Fifth Report, Part I: NationalOverview. Centre for Science and Environment, New Delhi.

EERE – Energy Efficiency and Renewable Energy, U.S. Dept. of Energy, 2013. CleanCities Alternative Fuel Price Report. Jan. 2005 to Apr. 2013. Golden, Colo.: U.S.Dept. of Energy, Energy Efficiency and Renewable Energy, Clean Cities. AccessedJune 2014. ⟨http://www.eere.energy.gov/⟩.

EIA – U.S. Energy Information Administration, 2013. US Natural Gas Price, CitigatePrice, 2013, Annual [Dataset in website]. Washington, D.C.: U.S. Department ofEnergy: Energy Information Administration. Accessed June 2014. ⟨http://www.eia.gov⟩.

GNCTD – Govt. of NCT of Delhi, 2012. Statistical Abstract of Delhi 2012. Delhi, India:Directorate of Economics and Statistics, Govt. of National Capital Territory ofDelhi. Downloaded June 2014. ⟨http://delhi.gov.in/wps/wcm/connect/doit/DelhiþGovt/DelhiþHome⟩.

Government of India, 2013. Monthly All-India Consumer Price Index Numbers forIndustrial Workers. Labour Bureau, Government of India, Data retrieved Octo-ber 2013 ⟨http://labourbureau.nic.in/main2.html⟩.

Government of India, 2009a. Minutes of the 68th Meeting of Central Sanctioningand Monitoring Committee (CSMC) on 13.2.2009, for sanctioning projects. D.O.No.K�14011/48/2006-UT (Pt.). Ministry of Urban Development, Government ofIndia. Downloaded June 2012. ⟨http://jnnurm.nic.in/⟩.

Government of India, 2009b Minutes of the 69th Meeting of Central Sanctioningand Monitoring Committee (CSMC) on 20.2.2009 and 21.2.2009, for sanctioningprojects. D.O.No.K�14011/48/2006-UT (Pt.). Ministry of Urban Development,Government of India. Downloaded June 2012. ⟨http://jnnurm.nic.in/⟩.

Government of India, 2009c. Minutes of the 70th Meeting of Central Sanctioningand Monitoring Committee (CSMC) of 26.2.2009, for sanctioning projects underUIG of JNNURM.K�14011/48/2006-UT (Pt.). Ministry of Urban Development,Government of India. Downloaded June 2012. ⟨http://jnnurm.nic.in/⟩.

Government of India, 2011. Minutes of the 95th Meeting of Central Sanctioning andMonitoring Committee (CSMC) on 15.3.2011, for sanctioning projects under UIGof JNNURM.K�14012/2(95)/2006-NURM. Ministry of Urban Development,Government of India. Downloaded June 2012. ⟨http://jnnurm.nic.in/⟩.

C. Krelling, M.G. Badami / Transport Policy 47 (2016) 178–188188

Government of India, 2012. Minutes of the 104th Meeting of Central Sanctioningand Monitoring Committee (CSMC) for sanctioning projects. K�14011/3/2012-UT-I. Ministry of Urban Development, Government of India. Downloaded June2012. ⟨http://jnnurm.nic.in/⟩.

Jalihal, S.A., Reddy, T.S., 2006. Assessment of the impact of improvement measureson air quality: case study of Delhi. J. Transp. Eng. 132 (6), 482–488. http://dx.doi.org/10.1061/(ASCE)0733�947X(2006)132:6(482).

Kathuria, V., 2005. Vehicular pollution control in Delhi: impact of compressednatural gas. Econ. Political Wkly. 40 (18), 1907–1915.

Kojima, M., Brandon, C., Shah, J., 2000. Improving Urban Air Quality in South Asiaby Reducing Emissions from Two-stroke Engine Vehicles. The World Bank,Washington, DC.

Kumar, N., Foster, A., 2009. Air quality interventions and spatial dynamics of airpollution in Delhi and its surroundings. Int. J. Environ. Waste Manag. 4 (1),85–111.

Lowell, D., Chernicoff, W., Lian, F., 2007. Fuel Cell Bus Life Cycle Cost Model: BaseCase & Future Scenario Analysis. U.S. Department of Transportation, Wa-shington, D.C..

Melendez, M., Taylor, J., Zuboy, J., Wayne, W.S., Smith, D., 2005. Emission testing ofWashington Metropolitan Area Transit Authority (WMATA) Natural Gas andDiesel Buses. National Renewable Energy Laboratory, U.S. Department of En-ergy, Golden, CO..

Municipal Corporation Ludhiana, 2013. Request for proposal (RFP) from bus op-erators for part financing & procurement, operation & maintenance of citybuses on specified routes under city bus services project in Ludhiana underJnNURM. Municipal Corporation Ludhiana, Ludhiana, India, Downloaded June25, 2013 ⟨http://mcludhiana.com/Attach%5CRfp%5CRFP_BusOperator_LCBSL_Final_14thMarch2013_LegallyVetted.pdf⟩.

MyPetrolPrice.com (2013). “Diesel Prices in Delhi” and “CNG Prices in Delhi”[website dataset]. Retrieved July 2013. ⟨http://www.mypetrolprice.com/⟩.

Narain, U. & Krupnick, A. (2007). The Impact of Delhi's CNG Program on Air Quality.RFF DP 07�06. Washington D.C.: Resources for the Future. Accessed July 2010.

Patankar, M., Patwardhan, A., 2006. The switch to CNG in two urban areas in India,in Nieuwenhuis,. In: Vergragt, Wells (Eds.), The Business of Sustainable Mobi-lity – From Vision to Reality. Greenleaf Publishing Ltd, Sheffield, U.K,pp. 171–186.

Posada, F., 2009. CNG Bus Emissions Roadmap: from Euro III to Euro VI. The In-ternational Council on Clean Transportation. Downloaded July 2013. ⟨http://

www.theicct.org/sites/default/files/publications/CNGbuses_dec09.pdf⟩.Reserve Bank of India, 2013. Handbook of Statistics on the Indian Economy (Ta-

ble 147). Reserve Bank of India. Data downloaded Oct. 2013. ⟨http://www.rbi.org.in/home.aspx⟩.

Reynolds, C.C.O., Kandlikar, M., 2008. Climate impacts of air quality policy:switching to a natural gas-fueled public transportation system in New Delhi.Environ. Sci. Technol. 42, 5860–5865.

Reynolds, C.C.O., Kandlikar, M., Badami, M., 2011. Determinants of PM and GHGemissions from natural gas-fueled auto-rickshaws in Delhi. Transp. Res. D 16,160–165.

Sen, A.K., Tiwari, G., Upadhyay, V., 2007. Should bus commuting be subsidized forproviding quality transport services? Sadhna 32 (4), 329–346, DownloadedAug. 2010 ⟨http://web.iitd.ac.in/�tripp/publications/pubgroup/transportation.htm⟩.

SIAM – Society of Indian Automobile Manufacturers, 2013. Fuel Specifications: In-dian Diesel Specification [Website]. Retrieved February 2013. ⟨http://118.67.250.203//scripts/Diesel.aspx⟩.

Supreme Court of India, 1998. Judgment in the matter of M.C. Mehta vs. Union ofIndia & Ors. –Writ Petition No. 939 of 1996. New Delhi: Supreme Court of India.July 28.

TERI – Tata Energy Research Institute, 2002. Pricing and Infrastructure Costing forSupply and Distribution of CNG and ULSD to the Transport Sector in Mumbai,India. New Delhi: TERI.

The Times of India, July 2013. Breakdowns mar low-floors' track record. Onlinearticle. ⟨http://timesofindia.indiatimes.com/city/delhi/Breakdowns-mar-low-floors-track-record/articleshow/20868560.cms?referral¼PM⟩ (accessed Feb-ruary 2014).

Wayne, W., Clark, N., Khan, A., Gautam, M., Thompson, G., Lyons, D., 2008. Regu-lated and non-regulated emissions and fuel economy from conventional diesel,hybrid-electric diesel, and natural gas transit buses. J. Transp. Res. Forum 47 (3),105–125.

WHO, 2014. Ambient (outdoor) air pollution database, by country and city. WorldHealth Organization, Geneva, Accessible at ⟨http://www.who.int/phe/health_topics/outdoorair/databases/cities/en/⟩.

WSA – Wilbur Smith Associates, 2008. Study on Traffic and Transportation Policiesand Strategies in Urban Areas in India-Final Report. Wilbur Smith Associates,Bangalore. Downloaded July 2009. ⟨http://urbanindia.nic.in/moud/programme/ut/final_Report.pdf⟩.