ww.sciencedirect.com

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1

Available online at w

ScienceDirect

http: / /www.elsevier .com/locate/biombioe

Laboratory based assessment of cookstoveperformance using energy and emissionparameters for North Indian cooking cycle

Pooja Arora a, Suresh Jain a,b,*, Kamna Sachdeva a

a Department of Natural Resources, TERI University, 10, Institutional Area, Vasant Kunj, New Delhi 110070, Indiab Department of Energy and Environment, TERI University, 10, Institutional Area, Vasant Kunj, New Delhi 110070,

India

a r t i c l e i n f o

Article history:

Received 15 December 2013

Received in revised form

9 July 2014

Accepted 13 July 2014

Available online

Keywords:

Specific energy consumption

Controlled cooking test

Biomass energy

Air emissions

Cooking cycle

Water boiling test

* Corresponding author. Department of NatuTel.: þ91 11 2612 2222; fax: þ91 11 2612 2874

E-mail addresses: sureshjain_in@yahoo.chttp://dx.doi.org/10.1016/j.biombioe.2014.07.0961-9534/© 2014 Elsevier Ltd. All rights rese

a b s t r a c t

The study was conducted to evaluate the performance of cookstoves using controlled

cooking test (CCT) with cooking cycles of Northern India, i.e., Uttar Pradesh (UP) and

Uttarakhand (UK). Cooking time, specific energy consumption (SEC), and emission factors

(EFs) of carbon monoxide (CO) and particulate matter (PM) were monitored for three

improved and one traditional cookstoves. Cooking was conducted by residents of the study

area. The findings from the study showed that the difference in thickness of roti baked in

UP and UK resulted in a significant change in emission and energy performance in all the

four cookstoves. The low-power inputs required for baking thinner roti resulted in higher

CO and PM emissions in case of CCT-UK. The results of CCT were also compared to Water

Boiling Test (WBT) for all the four cookstoves. The percent reductions in terms of energy

and emission parameters in the three improved cookstoves compared to traditional

cookstove were found to differ in CCT andWBT. Large variations were also observed during

the emission performance (40% decrease in CO EF) of the fan cookstove with change in fan

speed, which was usually unreported in previous studies. Overall, results show significant

influence of cooking cycles on cookstove performance, which was found to alter the

cookstove rankings. Therefore, the study thrusts upon the inclusion of user centric cook-

stove testing protocols in order to identify actual benefits for targeted rural communities.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

In most of the developing countries, biomass based cook-

stoves are a lifeline of any rural household, given the limited

financial resources. These bottom of the pyramid (BoP) energy

systems are being regularly redesigned and tested, in order to

ral Resources, TERI Unive.om, sureshj@teri.res.in (S012rved.

provide the underprivileged sections of the society, a healthy

and sustainable lifestyle [1]. Testing of cookstoves is one of the

most crucial steps in any cookstove dissemination program.

Cookstove testing is conducted in a laboratory, and/or in field

conditions, to assess the performance in terms of fuel savings

(energy efficiency) and environmental parameters, i.e., emis-

sion of air pollutants. Particulate matter (PM) and carbon

rsity, 10, Institutional Area, Vasant Kunj, New Delhi 110070, India.

. Jain).

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1212

monoxide (CO) due to their relevance in terms of health and

climatic impacts are the most commonly studied indoor pol-

lutants [2,3]. Themostwidely used laboratory test is theWater

Boiling Test (WBT), which is a simulation of actual cooking

style. The latest version of WBT (Version 4.1.2) provides a

methodology to test energy efficiency and emissions simul-

taneously, in order to understand the trade-offs that exist

between these two parameters [4]. Under the Household

Environment and Health (HEH) Project, two more testing

protocol were developed apart from WBT, namely, Controlled

Cooking Test (CCT), and Kitchen Performance Test (KPT). CCT

and KPT give a better representation of cookstove perfor-

mance because of the direct involvement of the beneficiaries.

Some CCT and KPT based studies have evidently shown a

weak link between the actual and simulated cooking condi-

tions [5e10]. These studies have relied on either fuel savings

or emission reduction to assess the cookstove designs. How-

ever, for a better assessment, energy and emission parame-

ters should be monitored concomitantly. Besides the

understanding of these factors, region specific cooking re-

gimes are also crucial part for the cookstove testing [11]. WBT

has also been used to associate the cookstove performance

with different variables such as fuel and vessel characteristics

[12e14]. These variables are representative of varied field

scenarios, and the individual impact of these variables can be

better understood under controlled laboratory conditions. The

concept of ‘burns cycle’ or ‘cooking cycle’ introduced by

Johnson et al. [15] and has been considered as a critical factor

in assessing the actual performance of cookstoves. CCT can

prove to be a better method for testing cookstoves as it can

easily capture the intricate details of cooking cycles, which

can help both stove designers and researchers engaged in

development of cookstove testing protocols. In the best of the

author's knowledge, no study has been carried out for Indian

conditions, where energy and emissions from cookstoves

have been reported using regional cooking cycles except one

study conducted in Tamilnadu by MacCarty et al. [16].

Therefore, this study proposes to assess cookstove perfor-

mance in terms of energy and emissions based on ‘Indian

cooking cycles. These results would be able to measure the

variations in cookstove performance using CCT protocol,

owing to the difference in cooking cycles existing in two

different regions of northern India i.e. Uttar Pradesh (UP) and

Uttarakhand (UK).

Fig. 1 e Photographs of the cookstoves tested. (a) Philips

forced draft (PF), (b) Philips natural draft (PN), (c) Envirofit

front feed (EN), (d) Traditional brick-mud cookstove (TR).

2. Methodology

2.1. Cookstoves tested, and test fuels used

The cookstoves selected in the study are among some of the

popular designs available in the Indian markets as shown in

Fig. 1. The four cookstoves differed from each other in terms of

the build material, size of the combustion chamber, fuel

feeding style and source of air supply (fan/free convection).

The top feed forced draft cookstove manufactured by Philips

India Limited, was selected under the advanced cookstove

category. The top feed natural draft (manufactured by Philips)

and front feed natural draft (manufactured by Envirofit In-

ternational Limited) were selected under improved cookstove

category. However, for the ease of discussion the above

mentioned cookstoves would be denoted as “improved cook-

stove” in the rest of the paper. The improved cookstoves were

tested against a commonly used traditional brick-mud cook-

stove (front feed mud cookstove, TR) design used in Northern

India. The TR cookstove was fabricated in the laboratory on a

metal sheet in order tomake it portable for the ease of testing.

Although in some areas within the study region, double pot

cookstoves are also used. This study, however, is limited to

single pot cookstoves.

All the tests were conducted using one wood, i.e., Acacia

nilotica (commonly known as Keeker), which is the most

commonly used wood for cooking at testing site. It was also

observed during the survey thatA. niloticawas also found to be

common cooking fuel in UP, in combination with other

biomass fuels such as cow dung and crop residue. During the

survey conducted in UK, the frequently used wood has been

oak (Quercus leucotrichophora and Quercus semecarpifolia); how-

ever, past studies have shown that the performance of cook-

stove more depends on cookstove designs rather than wood

types, except hardwoods and softwoods [17]. The wood size

used in each cookstove was set according to the manufac-

turer's product manuals. The wood size for the two top feed

cookstoves (PF and PN cookstoves) was 2 � 3 � 10 cm and

1.5� 1.5� 25 cm for the EN cookstove and 3� 3� 25 cm for the

TR cookstove. The wood moisture mass fraction during the

testing period was 8e14% (wet basis) measured using oven

method, which is also the case in the field. The dry seasons in

northern India can last upto nine months including winter

and summer during thewhich the fuelmoisture can vary from

�5% to �15%. The wood moisture can vary under different

relative humidity and temperature conditions also demon-

strated by Simpson and TenWolde [18]. The calorific values for

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1 213

wood and charcoal were calculated using bomb calorimeter

(Widsons Scientific Works, India) and were found to be

20.8 MJ kg�1 and 32.0 MJ kg�1 respectively.

2.2. Survey for food consumption and cooking cycle

Number of food items and their composition in a single meal

was set according to the most widely used cooking style in

Northern part of India. The cooking styles and food itemswere

identified by the help of a survey conducted in two states of

India, UK and UP. The survey was conducted in two villages in

the UK, Thalan village (30� 42� 56Ø N/78� 29� 06Ø E) in the

district of Uttarkashi and AamKheri (29� 47� 35Ø N/77� 52� 46Ø

E) village in Roorkee lying on the plains of UK. In Western UP,

Asoda (28� 45� 16Ø N/77� 46� 21ØE) and Dhaulana (28� 42� 43Ø

N/77� 39� 15ØE) villages in Ghaziabad district and Isagarh (25�

27� 28Ø N/78� 31� 16Ø E)and Bakhtar (25� 15� 03Ø N/79� 06� 37Ø

E) villages in Jhansi district were surveyed. While, in Eastern

UP, the survey was conducted to the village named Balia (25�

45� 31Ø N/84� 08� 55Ø E). The sampling was conducted using

stratified random sampling with 50e100 households within

each village. The survey was conducted using a semi-

Fig. 2 e Schematic diagram showing steps f

structured questionnaire which comprised of questions

related to the primary food items, its quantity and frequency

and cooking sequence of each food item.

2.3. Test protocols

The protocol used to assess the cookstove performance is

commonly known as CCT Version 2.0 [19], developed under

HEH Program, Shell Foundation. CCT allows comparing

cookstoves assuming their adaptability to the predominant

meals cooked in the community to be benefited. Themeal was

prepared by two female cooks who were residents of UP and

UK, respectively. The cook was appointed only after it was

made sure that they followed a similar cooking style as

observed within the study area. Weight of each food item, i.e.

raw rice and pulse along with water and cooking vessel and

wheat flour was recorded at the end of the test as presented in

Fig. 2. Sampling was begun as soon as the vessel was put over

the cookstove, and initial time was recorded. Ignition style

was kept constant for all the tests using wood chips prepared

from the same lot of wood used as fuel. The charcoal formed

ollowed during controlled cooking test.

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1214

was also weighed along with the cookstove and was used for

next cooking phase.

The cookstoves were also tested usingWBT Version 4.12 in

order to compare their performance with results obtained

with CCT. WBT is a simulation of actual cooking cycles,

developed principally to study the effect of design on the

performance of cookstoves. WBT primarily consists of two

phases, a high-power phase and low power phase as pre-

sented in Fig. 3. The high-power phase is performed in two

steps, i.e., cold start and hot start, where the latter highlights

the capacity of the cookstove to retain heat. The low-power

phase involves heating the water at a temperature 3 �Cbelow the local boiling point (BP) for 45 min, and represents

the low-power cooking tasks such as simmering of rice and

pulses. WBT was used to calculate energy parameters, i.e.,

thermal efficiency (TE), specific energy consumption (SEC),

firepower (FP)/power input and boiling time (BT) whereas, for

CCT, SEC, FP/power input, and cooking time (CT) were evalu-

ated. PM and CO sampling was carried out for all the three

phases of CCT as well as WBT. CO reading was taken every

30 s, depending on the high variability in emissions during the

cooking process. Emission factors (EFs) were calculated using

Fig. 3 e Schematic diagram showing step

the exhaust flow method of determining the equivalent wood

consumed, by normalizing it with moisture content and

charcoal left. All tests were performed in five replicates and in

case of CCTs, additional time was allotted to train the cooks to

make them aware with the cookstove operation procedure.

2.4. Sampling set-up and emission calculation

The testing facility consisted of a hood-duct system was used

for estimation of air emissions as shown in Fig. 4. The size of

the hood was enough to accommodate the cookstoves man-

ufactured for domestic cooking purposes. The PM sampling

point was kept ~9 times the duct diameter from the closest

point of disturbance and a CO sampling point was kept 1.5

feet above the PM sampling point to avoid the affect on each

other. A flue gas analyzer (Model E100 E instruments, USA)

based on electrochemical sensor was used for CO sampling.

PM sampling setup consisted of silicon tubings attached to a

filter holder carrying a 47 mm quartz filter (Q-MA, with pore

size 2 mm) and the PM deposited onto the filter consisted of

total particulate matter (TPM). The blower was used to create

a negative pressure within the duct, quite high to extract all

s followed during water boiling test.

Fig. 4 e Schematic diagram of the sampling setup.

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1 215

flue gas emitted and sufficiently low to avoid quenching of

combustion temperature inside the cookstove. The dilution

ratio (DR) was calculated on the basis of combustion and duct

temperatures, and was found to be 24 in the present study.

The value was close to DRs reported by Roden et al. [4] in

laboratory-based results used in cookstove performance. The

PM sampling was conducted under strictly controlled iso-

kinetic conditions. The flue gas velocity through the duct

was calculated using a pitot tube, by measuring the pressure

and temperature concomitantly. A flow rate of the sampling

pump was set according to the velocity measured (~3.5 ms�1)

with pitot tube and nozzle size of the sampling probe. The EFs

were calculated using the exhaust flow method recom-

mended in WBT (Version 4.1.2) which has also been used in a

study conducted by Venkatraman and Rao [20]. The hood was

covered from 3 sides with metal sheets to prevent spillage of

smoke. The test site being located in a clean locality, there

was no considerable background concentration of the pol-

lutants measured for this study. Please refer Arora et al. [17]

for more details on the sampling procedure used in the pre-

sent study.

2.5. Statistical analysis

Two-way anova is used with two factors presented in rows

and columns in a factorial design. It tests the difference in

means using three hypotheses, which consist of two main

effects (difference in means between cookstoves and cooking

phases, respectively) and an interaction (difference in means

between types of cookstove and cooking phases) [21]. Factorial

design consisted of two factors i.e. cookstove types (four levels

- TR, PF, PN and EN) and cooking phases (three levels e pulse,

rice and roti making). Further, Students' t-test was conducted

at a significance level of 0.05 in order to identify pairs with

statistically significant difference in means with the help of

Bonferroni correction method, where a ¼ a/c (c is the number

of paired means comparisons) [22].

3. Results and discussion

3.1. Survey results

Data collected from individual villages of the UP, and UK

during the primary survey showed that ~64% of the house-

holds have a similar cooking sequence for different food items

in a single meal, i.e. either pulseericeeroti or pul-

seevegetableeroti (roti is a common name for Indian wheat

bread). Rice and vegetable were found to be cooked alterna-

tively in a singlemeal throughout theweek. Rotis were cooked

mostly at the end because of their tendency to lose moisture.

In the present study, pulseericeeroti sequence was used in

the experiments. The vegetable preparation was not used for

the experiments because oil use in vegetable preparation may

lead to error in PM quantification and its chemical charac-

terization due to oil deposition. Ricewas found to be dominant

in the villages of Eastern UP, and the roti was dominant in

villages of Western UP and UK. However, the average quantity

of each pulse and rice was found to be ~250 g and ~300 g,

respectively, in all villages of UP and UK. At the same time, the

amount of wheat consumption was found to be 750 g in UP

while it was 500 g in case of UK. It was also found that more

than 70%households prepared rotis with thickness of 4e5mm

in villages of UP compared to ~2 mm in villages of UK. The

quantity of each food item (dry basis) in single meal per

household (kg d�1) was validated using food consumption and

average household size (~5 people per household for both UK

and UP) data from National Sample Survey Organization

(NSSO) in 2009e2010 for rural population in the form of kg d�1

in UK and UP, separately [23]. The data collected through a

survey for the quantity of pulse and wheat consumed in the

two states were comparable with NSSO data. The study

focussed on the most common cooking cycle; therefore, slight

variations in cooking styles were not accounted (Please see

Supplementary Information (SI), Table-S1 for descriptive sta-

tistics of survey results).

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1216

Vessel dimensions and material were kept the same as

noted over the field area. Therefore, aluminum flat bottomed

vessels were selected, and size was selected on the basis of

quantity of each food item. The water required for each food

item was taken according to the data procured from the field.

The pulse and rice were cooked by adding spices (in case of

pulses only) before putting onto the cookstove andwas heated

until it came to boil. The pulse was cooked for a few more

minutes until a certain texture was observed, indicating the

end of cooking. Preparation of roti involved making of dough

and dividing it into equal parts. Each part was subsequently

rolled upon and flattened into a disc-shaped roti and put on

the hot pan/tava. Roti was then partially baked from two sides

on a low flame and was then baked on high flame leading to

puffing up of the roti.

3.2. Cooking time

The cooking time for each of the three phases, i.e. pulse, rice

and roti were compared using two CCTs for all cookstoves as

presented in Fig. 5 (a,b). Total time required to complete the

cooking cycle was higher in CCT-UP compared to CCT-UK in

all the cookstoves. Among all the cookstoves, EN cookstove

was most sensitive with 22% increase in cooking time during

CCT-UP compared to CCT-UK. The increase in total cooking

time during CCT-UP was due to extra time spent in pulse

-

5

10

15

20

25

30

35

40

45

Puls

e

Ric

e

Rot

i

Puls

e

Ric

e

Rot

i

Puls

e

Ric

e

Rot

i

Puls

e

Ric

e

Rot

i

TR PF PN EN

Coo

king

tim

e (m

in)

(a) CCT-UK CCT-UP

0

2

4

6

8

10

12

14

16

Puls

e

Ric

e

Rot

i

Puls

e

Ric

e

Rot

i

Puls

e

Ric

e

Rot

i

Puls

e

Ric

e

Rot

i

TR PF PN EN

SEC

(M

Jkg-1

)

(c)

Fig. 5 e Variation in (a) Phase-wise cooking time, (b) Total cooki

and CCT-UP in the four cookstoves.

cooking and roti making phases. During pulse cooking phase

in CCT-UP, the cook preferred to boil the pulse-water mixture

for a longer time compared to the cook in case of CCT-UK.

The roti making phase required more time to complete the

task in CCT-UP compared to CCT-UK due to change in

thickness of roti which took longer time to get baked

compared to the thinner roti. The improvement in terms of

savings in cooking time compared to TR cookstove was only

evident in PF cookstove with 16% and 10% reduction in total

cooking time (p < 0.0083) during CCT-UK and CCT-UP

respectively, while PN and EN cookstove showed no signifi-

cant improvement over TR cookstove. The meager difference

in total cooking time in PN and EN cookstoves was due to

18e37% increase in time spent during roti making phase

which impacted the net savings in cooking time based on the

three cooking phases. The advantage with the TR cookstove

was found to be the process of puffing up of roti which was

done on the glowing charcoal inside the combustion cham-

ber. Therefore, it was possible for the cooks to bake two roti'ssimultaneously, thereby reducing overall cooking time. In

case of all other cookstoves, this step was carried out on the

pan itself because of the difference in geometry of the com-

bustion chamber.

(Please see Supplementary Information (SI), Tables S2 and

S6 for the results of Two-way anova with replication and

Student's t-test and values for total cooking time).

0

10

20

30

40

50

60

70

80

90

TR PF PN EN

Tota

l coo

king

tim

e (m

in)

(b) CCT-UK CCT-UP

0

1

2

3

4

5

6

7

8

9

10

TR PF PN EN

Ave

rage

SE

C (

MJk

g- 1)

Cookstoves

(d)

ng time, (c) Phase-wise SEC and (d) Average SEC in CCT-UK

Table 1 e Power inputs (Watts) during different phases of combustion in WBT, CCT-UK and CCT-UP.

Protocol Type ofcookstoves

Phase 1 Phase 2 Phase 3 Mean of three combustion phases ± SD

Mean ± SD Mean ± SD Mean ± SD

WBT Cold Start Hot Start Simmer

TR 4200 ± 389 6074 ± 809 4423 ± 525 4899 ± 1023

PF 4581 ± 288 4882 ± 441 2166 ± 174 3876 ± 1489

PN 3442 ± 394 3434 ± 413 2299 ± 236 3058 ± 658

EN 3711 ± 437 3591 ± 300 2226 ± 114 3176 ± 825

CCT-UK Pulse Rice Roti

TR 5249 ± 382 6689 ± 606 3223 ± 182 5053 ± 1741

PF 3568 ± 428 3609 ± 360 1458 ± 65 2878 ± 1230

PN 3048 ± 188 3017 ± 239 2537 ± 169 2867 ± 286

EN 2461 ± 161 3951 ± 392 4026 ± 214 3479 ± 883

CCT-UP Pulse Rice Roti

TR 4053 ± 190 5729 ± 84 5372 ± 154 5051 ± 883

PF 2986 ± 272 3185 ± 294 3408 ± 30 3193 ± 211

PN 3057 ± 131 3995 ± 416 2622 ± 78 3225 ± 702

EN 3150 ± 188 4721 ± 116 3037 ± 233 3636 ± 941

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1 217

3.3. Specific energy consumption

Fig. 5(c, d) shows SEC variation in all cookstoves during the

three cooking phases in CCT-UK and CCT-UP. The average SEC

of all the three cooking phases was 31e47% higher during

CCT-UK compared to CCT-UP in all the cookstoves with only

10e26% decrease in power inputs. Summary of power inputs

during each phase of cooking and average of three of phases in

presented in Table 1. The difference in SEC can be attributed to

the change in combustion conditions during the two cooking

cycles. It was observed that the increase in cooking time

during the first phase i.e. pulse cooking in CCT-UP led to a

decrease in SEC in the ensuing rice cooking phase which in-

dicates wastage of fuel during CCT-UK. This signifies the

impact of user behavior on the cookstove performance when

choice is between shorter cooking time and lesser fuel con-

sumption. Another impact of difference in cooking style on

cookstove performance was observed during roti making

phase. In case of the CCT-UK, SEC was higher during roti

making due to higher charcoal consumption, owing to the

smoldering conditions required to maintain low temperature

for baking the thinner roti. The increase in SEC during low

power cooking is also due to increase in heat loss to the sur-

rounding environment as a consequence of improper insu-

lation of combustion chamber or low fuel to air ratios.

Contrary to CCT-UK, the cooking cycle in CCT-UP was domi-

nated by flaming conditions which was required to bake a

comparatively thick roti (Please see SI, Tables S-3 and S-6 for

the results of Two-way anovawith replication and Student's t-test and values of average SEC).

3.4. CO and PM EFs

Fig. 6(aed) shows the variation in CO and PM emissions in

CCT-UK and CCT-UP during all the three phases of cooking

cycle. The average CO and PM EFs were higher during CCT-UK

compared to CCT-UP in all the cookstoves. An increase of

~40% in CO EFs in CCT-UK compared to CCT-UP was observed

in case of PF cookstove and was found to be statistically

significant (p ¼ 7.0 � 10�5). Whereas an increase of ~49% in

average PM EFs in CCT-UK compared to CCT-UP was observed

in case of PF and TR cookstoves. The difference can be

attributed to the smoldering conditions required for baking

thinner roti in CCT-UKwhich resulted in higher CO and PM EF.

However, in CCT-UP high power input was required to bake

comparatively thick roti which was achieved by maintaining

constant flaming phase. The increase in CO emissions during

smoldering is due to the enhanced charcoal oxidation owing

to more oxygen delivered on its surface with the help of

external air supply. The high carbon and low oxygen and

hydrogen content of charcoal thus results in high CO emis-

sions [24,25]. Similarly, high PM emissions during the roti

making were due to smoldering conditions, which lead to the

formation of high mass particles emitted in the form of tar

balls, which have been found to be composed of brown carbon

species in some studies [26,27]. The improvement over TR

cookstove in terms of CO EFs resulted in poorest performance

of PF cookstove which showed negligible reduction in CO EFs

compared to ~28 and ~40% reduction in PN and EN cookstoves

respectively. In case of CCT-UP the reduction in CO EFs was

~34% in PF and EN cookstoves, while PN cookstove showed no

considerable improvement over TR cookstove. Reduction in

PM EFs in the three improved cookstoves was ~54% in case of

CCT-UK, while in CCT-UP it was ~40% in PF and EN cookstove

however there was negligible improvement in case of PN

cookstove. The results indicate that roti making phase

emitted high CO and PM EFs compared with the pulse and rice

cooking phases. In addition, the difference in thickness of roti

was found to introduce a change in the two cooking cycles,

specifically in case of TR and PF cookstoves. (Please see SI,

Tables S4eS6 for the results of Two-way anova with replica-

tion and Student's t-test and values of average CO and PM EF).

The performance of the four cookstoves based on the total

cooking time, SEC and the emission parameters was used to

rank the cookstoves. Cookstove ranking based on total cook-

ing time during CCT-UK resulted in an order where

PF < EN < PN < TR; however, the difference in total cooking

time among PF, PN and EN cookstoves was not statistically

-

1

2

3

4

5

6

7

Puls

eR

ice

Rot

iPu

lse

Ric

eR

oti

Puls

eR

ice

Rot

iPu

lse

Ric

eR

oti

TR PF PN EN

CO

EF

(gM

J-1)

(a) CCT-UK CCT-UP

-

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Puls

eR

ice

Rot

iPu

lse

Ric

eR

oti

Puls

eR

ice

Rot

iPu

lse

Ric

eR

oti

TR PF PN EN

PM E

F (g

MJ-1

)

Cookstove & cooking phases

(c)

0

1

1

2

2

3

3

TR PF PN EN

Ave

rage

CO

EF

(gM

J-1)

(b) CCT-UK CCT-UP

0.00

0.10

0.20

0.30

0.40

0.50

0.60

TR PF PN EN

Ave

rage

PM

EF

(gM

J- 1)

Cookstoves

(d)

Fig. 6 e Variation in (a) Phase-wise CO EF, (b) Average CO EF, (c) Phase-wise PM EFs and (d) Average PM EF in CCT-UK and

CCT-UP in the four cookstoves.

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1218

significant (p > 0.0083). On the other hand, the cookstove

ranking was observed to be PF < TR < PN < EN during CCT-UP.

The difference was statistically significance in case of all

cookstove combinations (p < 0.0083) except TR-PN cookstove.

Comparison based on average SEC during CCT-UK and CCT-

UP, resulted in ranking of the cookstoves where

PF < PN < EN < TR, which implies that change in cooking cycle

did not impact the relative energy consumption in all the

cookstoves. However, the average CO and PM EFs obtained

during CCT-UK resulted in cookstove rankings where

EN < PN < PF < TR. Whereas in case of CCT-UP cookstove

rankings based on CO and PM EFs followed an order where

EN < PF < PN < TR and EN < PN < PF < TR respectively. The

rankings were found to change under the influence of each

parameter; however, the effect of change in test protocol was

evident only in case of emissions. The results clearly

demonstrate the importance of simultaneous measurement

of energy and emission parameters in order assess the actual

performance of cookstoves, which has also been reported by

Arora et al. [17,28]. There is also a need to formulate a ranking

methodology which reflects the combined effect of all the

parameters at the same time which will again depend on the

relative importance of the each of themeasured parameter i.e.

energy and emissions.

3.5. Relationship between CCT and WBT

Performance of all four cookstoves was compared, with re-

sults obtained from both CCTs and WBT as shown in

Fig. 7(aed). In terms of time savings, it was found that the

reduction in boiling time in WBT was ~58%, ~48% and ~32%

in PF, PN and EN cookstoves, respectively, when compared to

TR cookstove. However, the reduction in cooking time was in

the range of 4e16% in all the three cookstoves in CCT-UK and

CCT-UP. The reason for lower time savings in case of CCTs

compared to WBT was due to roti making phase. In the TR

cookstove, puffing of roti was done on charcoal bed, and

simultaneously, the other roti was baked on the tava/pan,

thus saving the time compared to the advanced cookstoves.

At the same time, the ranking of the three cookstoves

remained same for all the three tests in terms of time

savings.

0%

10%

20%

30%

40%

50%

60%

70%

PF PN EN

% r

educ

tion

com

pare

d to

TR

coo

ksto

ve

Cookstoves

(b) SEC

-20%

-10%

0%

10%

20%

30%

40%

50%

60%

70%

PF PN EN

% r

educ

tion

com

pare

d to

TR

coo

ksto

ve

(a) Cooking time

CCT-UK CCT-UP WBT

-10%

0%

10%

20%

30%

40%

50%

60%

70%

PF PN EN

% r

educ

tion

com

pare

d to

TR

coo

ksto

ve

(c) CO EF

0%

10%

20%

30%

40%

50%

60%

70%

PF PN EN

% r

educ

tion

com

pare

d to

TR

coo

ksto

ve

Cookstoves

(d) PM EF

Fig. 7 e Percent reduction in Cooking time, SEC, CO EF and PM EF in CCT-UK, CCT-UP and WBT.

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1 219

The reduction in average SEC was ~49% in all three cook-

stoves during WBT which was comparable to the reduction

observed during CCT-UK andCCT-UP. However the reductions

were similar only in case PF cookstove, whereas in PN and EN

cookstoves the reductionswere overestimated byWBT. This is

indicative of higher energy consumption (high power input)

and losses in PN and EN cookstoves during both the CCTs

compared to WBT. In PN and EN cookstoves, inadequate

insulation of the combustion chambers resulted in high

cookstove body temperatures which were found to be 65,

70 180 and 150 �C for TR, PF, PN and EN cookstoves, respec-

tively. However, the ranking of cookstoves was again alike for

all the three cookstoves during WBT and CCTs, as observed in

terms of absolute SEC for all the cookstoves. Ranking of the

cookstoves was also found to be similar to the rankings based

on SEC when they were compared based on TE of the three

cookstoves calculated during WBT. The maximum

temperatures attained can also be related to the user safety

where PN and EN cookstoves can pose a potential risk for skin

burns especially for children.

Overall, the reductions in CO and PM EFs compared to TR

cookstove in CCT-UP and CCT-UK were underestimated by

WBT. However, due to lack of consistent observations, the

relationship between CCT and WBT could not be established.

Similarly, reductions in PM emissions compared to TR cook-

stove were underestimated byWBT compared to CCT-UK. The

pattern of emission reductions showed that CCT-UP andWBT

were comparable in terms of measured emission parameters,

due to related combustion conditions and power inputs;

however, the association was again not consistent in all the

cookstoves.

The cooking cycle during the pulse and rice preparation

was found to be equivalent to high-power phase in WBT. It is

worth to highlight that once the pulseewater or rice-water

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1220

mixture reached the BP, the cook in both the CCTs kept boiling

the mixture at the same feeding rate for ~5 and ~10 min in

CCT-UK and CCT-UP, respectively. WBT states the low power

to be a simulation of simmer phase in pulse and rice cooking.

However, due to small food quantity, pulse and rice cooking

were not found to exhibit any considerable simmer phase as

stated by WBT. The simulation would be more representative

to the region where staple diet is rice leading to higher food

quantity, which might lead longer simmer phases. It was

observed that the North Indian cooking cycle is dominated by

high power in case of pulse, rice and roti (in case of thick roti).

In the present study, the average of CO EFs in all cookstoves

were found to be ~55% lower than the values 5.1 g MJ�1 re-

ported by Roden et al. [6] in a field study. Whereas, CO EFs of

the present study were comparable to CO EFs reported by

Jetter and Kariher [29], Roden et al. [10] and McCarty et al. [5]

estimated using WBT. Whereas, PM EFs were found to be

closer to field results reported by Roden et al. [6] (Please see SI,

Table S7 for comparison of present study CO and PM EFs with

values reported in the literature).

3.6. Effect of fan speed in PF cookstove

The PF cookstove was operated at the highest fan speed dur-

ing all three phases of WBT and CCTs. However, PF cookstove

was also operated at minimum air supply where, TE was

found to increase by ~15%; whereas, the CO EF decreased by

~42%. Thus, the results show that the extent to which the

amount of air supplied in forced draft cookstoves vary the

performance of a cookstove that in turn dependent upon the

discretion of the tester or the end user to set at the desired

level. TE achieved in the present study was 20e30% lower

compared to a recent study carried out by Jetter and Kariher

[29] and Raman et al. [30]. The difference in TE may be due to

the effect of air supply levels, which were not reported in their

results as well as the difference in surface area to the volume

ratio of wood pieces (4 compared to 2 in the present study)

used in PF cookstove conducted by Raman et al. [30] were 50%

higher than the present study. The higher surface area to

volume ratios helps to accelerate the combustion process and

efficient heat transfer [31].

4. Conclusion

The aim of the study was to measure the effect of change in

cooking cycle on cookstove performance, in terms of energy

and emission parameters. The results from this study indicate

the origin of variation in performance indicators due to dif-

ference in cooking cycles in a particular region of India. In

absolute terms, the total cooking time did not show any sig-

nificant difference with change in cooking cycle, however, the

performance of cookstoves in terms of SEC and CO and PM EFs

was found to vary with change in cooking cycles. The major

difference found in the cooking cycle of the UP andUKwas the

thickness of the roti. The thinner roti in UK was found to in-

crease average SEC, CO and PM EFs compared to the CCT-UP. It

was also observed that the roti making task takes more time

for all three “improved” cookstoves and also resulted in higher

CO emissions in forced draft cookstove in comparison to the

TR cookstove because of their inability to cater roti making

style. There were inconsistencies when performance of

“improved” cookstoves was compared to the traditional

cookstove. WBTwas found to underestimate the performance

at several instances, especially in terms of emissions.

Conclusively, even slight variation in power levels at which

actual and simulated cookingwas found to alter both absolute

and relative performance of cookstoves. Therefore, these

types of user centric studies focused on cooking cycles in

different regions can help in identifying the power levels

involved in specific cooking tasks. This will further benefit to

test cookstoves using only those power levels that are domi-

nant in certain regions. The results also indicated that after

considerable training, the users can show similarities in stove

operation style, thus making it easier to compare cookstoves

on their actual performance, based on specific cooking cycles.

Eventually, the testing results will help to enhance the

acceptability of new cookstoves among rural communities

with diverse cooking habits. These data sets can also be used

as an input for cookstove emission modeling studies in order

to create a regional emission database by using food and

cooking fuel consumption patterns of the targeted

communities.

Acknowledgment

The authors would like to thank Ministry of New and

Renewable Energy India (MNREeGrant No. NREF/TU/2011/10),

McArthur Foundation (Grant no. 09-92304-000-GSS) and Min-

istry of Human Resource Development (MHRD e Grant no. 16-

53/2009-DL), Government of India, for the financial support

provided during the course of this study. The authors grate-

fully acknowledge the cooperation of Mr. Neeraj Sharma from

Department of Natural Resources, TERI University for editorial

assistance. The authors also acknowledge the contribution of

Ms. Priyanka Chauhan in preparation of graphics.

Appendix A. Supplementary data

Supplementary data related to this article can be found at

http://dx.doi.org/10.1016/j.biombioe.2014.07.012.

r e f e r e n c e s

[1] Dhoble R, Bairiganjan S. Cooking practices and cookstovesfield insights: a pilot study of user experience withtraditional and improved cookstoves. Chennai: Institute forFinancial Management and Research (IFMR); 2009 November.p. 32.

[2] Naeher LP, Smith KR, Leaderer BP, Neufeld L, Mage DT.Carbon monoxide as a tracer for assessing exposures toparticulate matter in wood and gas cookstove householdsof highland Guatemala. Environ Sci Technol2001;35(3):575e81.

[3] Grieshop AP, Marshall JD, Kandlikar M. Health and climatebenefits of cookstove replacement options. Energy Policy2011;39(12):7530e42.

b i om a s s a n d b i o e n e r g y 6 9 ( 2 0 1 4 ) 2 1 1e2 2 1 221

[4] [Internet] Water boiling test (WBT) version 4.1.2Cookstoveemissions and efficiency in a controlled laboratory setting;2009 Oct [cited 2013 Oct 14]. Available from:, http://www.pciaonline.org/files/WBT4.1.2_0_0.pdf.

[5] MacCarty N, Still D, Ogle D. Fuel use and emissionsperformance of fifty cooking stoves in the laboratory andrelated benchmarks of performance. Energy Sustain Dev2010;14(3):161e71.

[6] Roden CA, Bond TC, Conway S, Benjamin A, Pinel O,MacCarty N, et al. Laboratory and field investigations ofparticulate and carbon monoxide emissions from traditionaland improved cookstoves. Atmos Environ2009;43(6):1170e81.

[7] Boy E, Bruce N, Smith KR, Hernandez R. Fuel efficiency of animproved wood-burning stove in rural Guatemala:implications for health, environment and development.Energy Sustain Dev 2000;4(2):23e31.

[8] Bailis R, Berrueta V, Chengappa C, Dutta K, Edwards R,Masera O, et al. Performance testing for monitoringimproved biomass stove interventions: experiences of thehousehold energy and health project. Energy Sustain Dev2007;11(2):57e70.

[9] Berrueta VM, Edwards RD, Masera OR. Energy performanceof wood-burning cookstoves in Michoacan, Mexico. RenewEnergy 2008;33(5):859e70.

[10] Roden CA, Bond TC, Conway S, Pinel ABO. Emission factorsand real-time optical properties of particles emitted fromtraditional wood burning cookstoves. Environ Sci Technol2006;40(21):6750e7.

[11] Lambe F, Atteridge A. Putting the cook before the stove: auser-centred approach to understanding household energydecision-making: a case study of Haryana State, northernIndia. Sweden: Stockholm Environment Institute; 2012 May.p. 27. Working paper no. 2012e03.

[12] Bhattacharya SC, Albina DO, Khaing AM. Effects of selectedparameters on performance and emission of biomass-firedcookstoves. Biomass Bioenergy 2002;23(5):387e95.

[13] Yuntenwi EAT, MacCarty N, Still D, Ertel J. Laboratory studyof the effects of moisture content on heat transfer andcombustion efficiency of three biomass cook stoves. EnergySustain Dev 2008;12(2):66e77.

[14] L'Orange C, DeFoort M, Willson B. Influence of testingparameters on biomass stove performance and developmentof an improved testing protocol. Energy Sustain Dev2012;16(1):3e12.

[15] Johnson M, Edwards R, Berrueta V, Masera O. Newapproaches to performance testing of improved cookstoves.Environ Sci Technol 2009;44(1):368e74.

[16] MacCarty N, Still D, Ogle D, Drouin T. Assessing cook stoveperformance: field and lab studies of three rocket stovescomparing the open fire and traditional stoves in TamilNadu, India on measures of time to cook, fuel use, total

emissions, and indoor air pollution. Oregon: AprovechoResearch Center; 2008 January. p. 18.

[17] Arora P, Jain S, Sachdeva K. Physical characterization ofparticulate matter emitted from wood combustion inimproved and traditional cookstoves. Energy Sustain Dev2013;17(5):497e503.

[18] Simpson W, TenWolde A. Physical properties and moisturerelations of wood. Wood handbook: wood as an engineeringmaterial. Madison, WI: USDA Forest Service, Forest ProductsLaboratory; 1999. p. 3.1e3.24. General technical report FPL;GTR-113.

[19] Bailis R. Controlled cooking test (CCT) version 2.0 [Internet].Household Energy and Health Programme. Shell Foundation;2004 Aug [cited 2013 Oct 14]. Available from:, http://www.pciaonline.org/files/CCT_Version_2.0_0.pdf.

[20] Venkataraman C, Rao GUM. Emission factors of carbonmonoxide and size-resolved aerosols from biofuelcombustion. Environ Sci Technol 2001;35(10):2100e7.

[21] Frank H, Althoen SC. Statistics: concepts and applications.New York: Cambridge university press; 1994.

[22] Hochberg Y. A sharper Bonferroni procedure for multipletests of significance. Biometrika 1988;75(4):800e2.

[23] [Internet] National Sample Survey OrganisationNSS 66thRound; 2011 [cited 2013 Nov 25]. Available from:, http://mospi.nic.in/Mospi_New/upload/NSS_Report_538.pdf.

[24] Ohlemiller TJ. Smoldering combustion propagation on solidwood. Fire Saf Sci 1991;3:565e74.

[25] Koppmann R, Czapiewski KV, Reid JS. A review of biomassburning emissions, part I: gaseous emissions of carbonmonoxide, methane, volatile organic compounds, andnitrogen containing compounds. Atmos Chem Phys Discuss2005;5(5):10455e516.

[26] Chakrabarty RK, Moosmuller H, Chen LW, Lewis K,Arnott WP, Mazzoleni C, et al. Brown carbon in tar balls fromsmoldering biomass combustion. Atmos Chem Phys2010;10(13):6363e70.

[27] Maruf Hossain AMM. Volatility and mixing states of ultrafineparticles from biomass burning. J Hazard Mater 2012;205-206(29):189e97.

[28] Arora P, Das P, Jain S, Kishore VVN. A laboratory basedcomparative study of Indian testing protocol and waterboiling test. Energy Sustain Dev 2014;21:81e8.

[29] Jetter JJ, Kariher P. Solid-fuel household cookstoves:characterization of performance and emissions. BiomassBioenergy 2009;33(2):294e5.

[30] Raman P, Murali J, Sakthivadivel D, Vigneswaran VS.Performance evaluation of three types of forced draftcookstoves using fuel wood and coconut shell. BiomassBioenergy 2013;49:333e40.

[31] Thomas PA, McAlpine RS. Fire in the forest. New York:Cambridge University Press; 2010.