grid Cooking Diaries Data - PDF document

1 A2EI Mini - grid Cooking Diaries Data A
A2EI Mini - grid Cooking Diaries Data Analysis Bezi & Mavota (Tanzania) A2EI Mini - grid Cooking Diaries Data Analysis Bezi & Mavota (Tanzania) A2EI smartmeter installed with an electric pressure cooker: Nelson Villema , 2020 February 2021 Authors: Tom Jones, Nigel Scott , Anna Clements (Gamos) MECS is funded by UK Aid through the Foreign, Commonwealth, and Development Office. It is a partnership between researchers, innovators, policy makers, and ESMAP drawing on their expertise and relevant work from around the world to co - construct new knowledge with practitioners and the private sector. It is led by Loughborough University, UK. The views expressed do not necessarily reflect the UK government’s official policies. A2EI Mini - grid Cooking Diaries Data Analysis Contents 1 Overview of data ................................ ................................ ................................ ................................ ............. 1 2 Energy consumptions ................................ ................................ ................................ ................................ ...... 5 2.1 Mix of fuels ................................ ................................ ................................ ................................ .............. 5 2.2 Per capita consumptions ................................ ................................ ................................ ......................... 7 2.3 Energy consumption by heating event ................................ ................................ ................................ . 10 2.4 Energy consumption per day ................................ ...............................

2 . ................................ .....
. ................................ ................ 15 2.5 EPC use during phase 2 ................................ ................................ ................................ ......................... 20 3 Meals cooked ................................ ................................ ................................ ................................ ................ 22 3.1 Food types cooked ................................ ................................ ................................ ................................ 22 3.2 Reheatin g food ................................ ................................ ................................ ................................ ...... 27 4 Cooking devices ................................ ................................ ................................ ................................ ............. 31 4.1 Detail on how participants cook ................................ ................................ ................................ ........... 31 4.2 Characteristics of different cooking devices ................................ ................................ ......................... 32 4.3 Fuel stacking ................................ ................................ ................................ ................................ .......... 33 4.4 Energy used by different stoves ................................ ................................ ................................ ............ 33 5 Time taken ................................ ................................ ................................ ................................ .................... 34 5.1 Time taken to cook food types ..........................

3 ...... ................................
...... ................................ ................................ ............. 34 5.2 Time taken to prepare meal ................................ ................................ ................................ ................. 36 5. 3 Time of day ................................ ................................ ................................ ................................ ............ 36 6 Water heating ................................ ................................ ................................ ................................ ............... 37 7 Conclusion ................................ ................................ ................................ ................................ ..................... 40 1 A2EI Mini - grid Cooking Diaries Data Analysis 1 Overview of data In order to investigate cooking habits among rural Tanzanian settlements, participants in two Tanzanian villages, Bezi and Mavota , were asked to keep cooking diaries for each cooked meal . There were 15 households who took part in Bezi and 16 in Mavota. The experiment was split into multiple phases; this report investigates the first two. In phase 1, participants were not asked to change their cooking habits, on ly to fill in the cooking diaries. In phase 2, households were given an electric pressure cooker (EPC) (and trained in how to use it) but not explicitly told to use it. The number of records obtained from each village and phase is shown in Table 1 . Paper records were kept by participants , copied into Kobo Toolbox by enumerators, and uploaded into an Excel worksheet. Every row in the worksheet contains the information for one record. E ach r

4 ecord could cover multiple heating eve
ecord could cover multiple heating events, e.g. an early morning record could include breakfast and preparing a snack, or preparing breakfast and lunch together . However, almost all data records covered only a single heating event ; there were 3088 records with at least 1 stated cooking event, but only 20 of the se records contained more than 1 purpose for their cooking . Table 1 Number of records from Bezi and Mavota in phases 1 and 2 Number of records Percent Bezi Phase 1 664 21.4 % Bezi Phase 2 735 23.7 % Mavota Phase 1 890 28.7 % Mavota Phase 2 807 26.1 % Total 3096 100 .0% T able 2 shows the heating event purposes for records that only recorded one event. In Bezi, dinner is the most common single heating event followed by lunch and then breakfast . I n Mavota, lunch is ju st as popular as dinner . Note that for each heating event, participants were asked to record details of each dish cooked and any water heated as part of the heating event. If water was heated on its own, and not as part of a meal, then it was logged as a ‘heating water’ event. T able 2 Number of heating events (single heating event records only) Bezi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 Breakfast 172 156 145 100 573 26.3% 21.2% 16.4% 12.5% 18.7% Lunch 205 265 362 361 1193 31.4% 36.1% 41.0% 45.3% 38.9% Dinner 269 313 376 336 1294 41.2% 42.6% 42.6% 42.2% 42.2% Food for baby 2 1 0 0 3 0.3% 0.1% 0.0% 0.0% 0.1% Heating water 2 0 0 0 2 0.3% 0.0% 0.0% 0.0% 0.1% Other 3 0 0 0 3 0.5% 0.0% 0.0% 0.0% 0.1% Total 653 735 883 79

5 7 3068 100.0% 100.0% 100.0% 10
7 3068 100.0% 100.0% 100.0% 100.0% 100.0% 2 A2EI Mini - grid Cooking Diaries Data Analysis Table 3 shows that in Bezi, more adults were cooked for than children whereas in Mavota, the opposite is true. In Bezi, the number of people cooked for was marginally higher in Phase 1 than in Phase 2 (Mann - Whitney p 0.007 ). However, the opposite trend can be seen in Mavota , where the number of people cooked for in Phase 2 was slightly higher (Mann - Whitney p 0.001). Differences across all three categories of adults, children, and total number of pe ople cooked for were all significant, with the exception of adults in Bezi 1 . The number of people cooked for at breakfasts was slightly lower than for lunches and dinners ( Table 4 ). Despite cooking for different demographics, both sets of households tended to prepare a mean of approximately 2 dishes per meal ( Table 5 ). There were margi nal differences in the number of meals cooked in Phase 1 and Phase 2 (Bezi Mann - Whitney p 0.001). Table 6 shows that lunch and dinner meals were both comprise roughly two dishes, but breakfasts tended to be simpler with a mean of only 1.3 dishes. Table 3 Average number of adults and children cooked for per record by Bezi and Mavota, phase s 1 and 2 Bezi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 Adults N 664 734 890 806 3094 Mean 3.8 3.7 2.3 2.3 3.0 Median 3 3 2 2 3 Std. Dev 1.7 1.3 1.0 0.8 1.4 Children N 663 734 890 806 3093 Mean 1.5 1.4 3.9 4.2 2.9 Median 2 1 4 4 3 Std. Dev 0.7 0.9 1.6 1.8 1.9 All people N 664 734 890 806 3094 Mean 5.4

6 5.1 6.2 6.5 5.8 Median 5 5
5.1 6.2 6.5 5.8 Median 5 5 6 6 5 Std. Dev 2.0 1.9 2.0 2.0 2.1 Table 4 Average number of adults and children cooked for per record by meal Breakfast Lunch Dinner Phase 1 Phase 2 Phase 1 Phase 2 Phase 1 Phase 2 Adults N 317 255 567 626 645 649 Mean 3.0 3.0 2.9 3.0 3.0 3.0 Median 3 3 2 3 2 3 Std. Dev 1.5 1.2 1.6 1.3 1.5 1.3 Children N 317 255 567 626 644 649 Mean 2.6 2.7 3.1 3.0 2.9 2.8 Median 2 2 3 3 3 3 Std. Dev 1.8 2.1 1.8 2.0 1.6 1.9 All people N 317 255 567 626 645 649 Mean 5.6 5.6 6.0 6.0 5.9 5.8 Median 5 5 5 6 5 5 Std. Dev 2.1 2.1 2.1 2.1 1.9 2.0 1 Bezi: children, MW p 001; Mavota: adults, MW p = 0.001; childr en, MW p 1. 3 A2EI Mini - grid Cooking Diaries Data Analysis Table 5 Average number of dishes cooked and water heating events per meal in Bezi and Mavota in phase s 1 and 2 (single heating event records only) Bezi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 Dishes cooked N 629 727 868 781 3005 Mean 1.9 1.8 1.8 1.8 1.8 Median 2 2 2 2 2 Std. Dev 0.7 0.6 0.5 0.5 0.6 Water heated N 244 186 104 54 588 Mean 1.1 1.0 1.0 1.0 1.0 Median 1 1 1 1 1 Std. Dev 0.2 0.1 0.1 0.0 0.2 Table 6 Average number of dishes cooked and water heating events per meal by meal (single heating event records only) Breakfast Lunch Dinner Dishes cooked N 512 1193 1294 Mean 1.3 1.9 2.0 Median 1 2 2 Std. Dev 0.5

7 0.5 0.5 Water heated N 368
0.5 0.5 Water heated N 368 51 166 Mean 1.0 1.0 1.0 Median 1 1 1 Std. Dev 0.2 0.1 0.2 Table 7 and Table 8 show the number of heating events per day recorded by each participant, and the allocation of those days to Phase 1 and Phase 2 . In Bezi, 47 % of heating events fell into Phase 1, and 53 % into Phase 2. On the other hand, 52% of heating events in Mavota were during into Phase 1, and 48% were in Phase 2. Note that for any given participant, the number of heating events recorded in a day varies. This reflects variations in heating behaviour e.g. water may n ot be heated every day, lunches may only be prepared at weekends, and certain meals may not be prepared if the household eats out . Note that there are quite a few gaps in the data (mostly in Bezi); this often represents periods where the main cook travelle d away from the home, usually for personal, health, or business matters. 4 A2EI Mini - grid Cooking Diaries Data Analysis Table 7 Number of heating events per day by participant s in Bezi Table 8 Number of heating events per day by participants in Mav ota 5 A2EI Mini - grid Cooking Diaries Data Analysis 2 Energy consumptions For each of the four dominant fuels, energy consumptions have been calculated from deduced fuel consumptions (based on the before and after readings e.g. weight of wood (kg)) and the calorific values given in Table 9 . Table 9 Calorific values and conversion efficiencies 2 Fuel Calorific value Density Wood 15.9 MJ/kg Charcoal 29.9 MJ/kg Kerosene 34.9 MJ/ltr 0.8 kg/ltr Electricity 3.6 MJ/kWh 2.1 Mix of fuels Figure 1 shows fuel usage in each pha

8 se and village . Charcoal is the most p
se and village . Charcoal is the most popular fuel being used in in over 90% of heating events. In Mavota , charcoal use stay ed relatively constant ; in both phases, it wa s used in 93% of heating events . I n Bezi, charcoal use increase d from 82 % in phase 1 to 9 3 % in phase 2 . It appears that this change mainly came from participants switching from wood to charcoal. It is worth noting that e ven with the introduction of the electric cookstoves in Phase 2 , electricity was seldomly used as a cooking fuel. Households likely avoide d using their electric cookstoves due to the cost of electricity when compared to their current fuel of choice. The total energy consumptions of each fuel are presented in Figure 2 which follows a similar pattern to Figure 1 . The only biggest difference between fuel use and energy consumption is that in phase 2 electricity is used in 5% of heating events but only accounts for 0.2% of the total energy consumed. T his shows the relative efficiency of the EPC when compared to traditional stoves. Figure 1 Fuel choice 3 2 Source: World Bank (BLG14 Cooking Costs by Fuel Type.xlsx) 3 N.B. data from Bezi captured more heating events in phase 2 than in phase 1, while data from Mavota captured more heating events in Phase 1 than in Phase 2 (see Table 1 ). 6 A2EI Mini - grid Cooking Diaries Data Analysis Figure 2 Energy content of fuels used Table 10 shows that participants rarely used multiple fuels in heating events for either Phase 1 or Phase 2 . However, despite electricity being used in only 5% of phase 2 heating events, it forms part of a fuel stack in 75% of these events . Of the small number of phase

9 2 events that used multiple fuels, over
2 events that used multiple fuels, over 90% of fuel stacks involved electricity ( Table 11 ) . It may be that households either cannot cook an entire meal using an EPC or cannot afford to use an EPC for the whole meal. Table 10 Number of fue ls used in single heating event Number of Fuels per heating event 4 Bezi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 0 29 17 5 0 51 4.4% 2.3% 0.6% 0.0% 1.6% 1 624 702 885 777 2988 94.0% 95.5% 99.4% 96.3% 96.5% 2 11 16 0 30 57 1.7% 2.2% 0.0% 3.7% 1.8% Total 664 735 890 807 3096 100% 100% 100% 100% 100% 4 This represents the number of fuels for which valid consumption data was provided, i.e. for the 51 records with zero heating fuels, the fuel was specified, but no consump tion data was available. 7 A2EI Mini - grid Cooking Diaries Data Analysis Table 11 Fuel choices Bezi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 Wood 52 3 47 51 153 8.2% 0.4% 5.3% 6.3% 5.0% Charcoal 511 650 838 715 2714 80.5% 90.5% 94.7% 88.6% 89.1% Kerosene 61 27 0 0 88 9.6% 3.8% 0.0% 0.0% 2.9% Electricity 0 22 0 11 33 0.0% 3.1% 0.0% 1.4% 1.1% Charcoal and wood 11 2 0 1 14 1.7% 0.3% 0.0% 0.1% 0.5% Charcoal and electricity 0 13 0 27 40 0.0% 1.8% 0.0% 3.3% 1.3% Wood and electricity 0 1 0 2 3 0.0% 0.1% 0.0% 0.2% 0.1% Total 635 718 885 807 3045 100.0% 100.0% 100.0% 100.0% 100.0% 2.2 Per capita consumptions Energy consumption depends on the number of people being cooked for. Therefor

10 e, p er capita energy consumptions have
e, p er capita energy consumptions have been calculated simply by dividing the energy consumption for a given heating event by the number of people that the meal was cooked for. Note tha t adults and children have been given an equal weighting when calculating per capita consumptions. In order to investigate the per capita energy consumption of each fuel, records that used multiple fuels had to be excluded because they only partially acco unt for a heating events energy consumption . As seen in Table 10 , this means that 57 of the 3045 records with a heating event are missing in the calcu lated per capita energy consumptions seen in Table 12 and Table 13 . Totals indicate that cooking with charcoal uses at least 20 times as much energy as cooking with electricity . It is hard to compare kerosene to another fuel considering it was only used in Bezi and usage drop ped significantly in phase 2. Likewise, wood is also difficult to compare to other fuels as it was only used in Bezi during phase 1 and only 1 household frequently used it in Mavota. Data from Bezi and Mavota were combined to explore what effect the number of people being cooked for had on the median average per capita energy consumption. Figure 3 and Figure 4 s how that if more people are cooked for, less energy is required to cook for ea ch individual , indicating that economies of scale can be achieved when using either charcoal or electricity. 8 A2EI Mini - grid Cooking Diaries Data Analysis Table 12 Per capita energy consumptions and number of people cooked for in Bezi – single fuels only Phase 1 Phase 2 Kerosene Charcoal Wood Electricity Kerosene Charcoal Wood Hou

11 sehold ID Median per capita consumpti
sehold ID Median per capita consumption (MJ) Median number of people cooked for Number of meals cooked Median per capita consumption (MJ) Median number of people cooked for Number of meals cooked Median per capita consumption (MJ) Median number of people cooked for Number of meals co oked Median per capita consumption (MJ) Median number of people cooked for Number of meals cooked Median per capita consumption (MJ) Median number of people cooked for Number of meals cooked Median per capita consumption (MJ) Median number of people cooke d for Number of meals cooked Median per capita consumption (MJ) Median number of people cooked for Number of meals cooked 1 7.4 6.0 1 4.3 7 35 0.5 3 4 6.7 3 40 2 4.3 3 41 14.8 3 3 0.2 4 4 5.5 4 64 3 4.8 5 39 1.8 12 1 0.0 7 2 7.2 5 35 4 4.5 6 54 4.3 4 1 0.0 14 1 2.9 7 52 1.0 9.0 1 5 5.5 4 33 11.7 3 29 6 5.5 7 7 5.1 7 34 5.2 7 45 5.6 7.0 2 7 3.6 3 55 5.2 7 1 0.2 3 8 4.8 3 45 8 3.7 7 38 7.7 7 7 3.8 8 65 9 1.1 5 33 2.5 6 12 5.2 5.5 2 1.5 5 39 10 6.0 6 12 6.9 4 44 11 2.4 6.5 38 0.6 5 5 0.1 7 2 4.3 4 38 12 0.6 3.5 8 4.3 4 48

12 1.3 3 1 0.7 4.0 1 3.4 4
1.3 3 1 0.7 4.0 1 3.4 4 52 13 7.9 4 1 3.3 5 31 2.7 6 35 14 4.3 5 44 5.5 5 51 15 0.2 5.5 18 4.0 5 24 0.4 6 1 0.2 5 24 1.4 5 15 Total 0.9 5 61 4.1 5 511 5.5 7 52 0.2 4 22 0.2 5 27 4.8 5 649 5.2 7 3 9 A2EI Mini - grid Cooking Diaries Data Analysis Table 13 Per capita energy consumptions and numbe r of people cooked for in Mavota – single fuels only Phase 1 Phase 2 Charcoal Wood Electricity Charcoal Wood Household ID Median per capita consumption (MJ) Median number of people cooked for Number of meals cooked Median per capita consumption (MJ) Median number of people cooked for Number of meals cooked Median per capita consumption (MJ) Median number of people cooked for Number of meals cooked Median per capita consumption (MJ) Median number of people cooked for Number of meals cooked Median per capita consumption (MJ) Median number of people cooked for Number of meals cooked 1 4.6 11 12 2.9 11 39 0.0 11 1 4.8 11 10 2.9 11 48 2 4.9 5 46 6.4 5 3 6.2 5 47 7.0 5 2 3 3.8 7 53 0.1 7 1 4.3 7 51 4 4.7 6 54 0.2 6 1 5.1 6 56 5 6.2 4 51 5.3 5 47 6 3.1 5 57 0.2 4 1 1.5 5 53 7 4.9 7 56 0.1 7 3 5.4 7 47 8 10.7 5 43 0.0 5 1 6.0 5 51

13 9 7.7 5 45
9 7.7 5 45 10 3.1 9 57 3.8 9 55 11 9.5 6 55 4.8 8 1 12 8.0 3 60 0.1 4 2 8.5 4 54 13 2.9 9 66 3.9 8 66 14 6.6 5 53 5.3 5 59 15 3.2 5 69 0.2 6 1 3.6 6 64 16 3.8 6 61 4.7 6 4 3.0 7 55 Total 4.7 6 838 3.6 11 47 0.1 6 11 4.6 6 715 2.9 11 50 10 A2EI Mini - grid Cooking Diaries Data Analysis Figure 3 Per capita charcoal energy consumption in phase 2 against number of people cooked for Figure 4 Per capita electricity energy consumption in phase 2 against number of people cooked for 2.3 Energy consumption by heating event Figure 5 and Figure 6 show that kerosene is more likely to be used for breakfast than for lunch or dinner, whereas the reverse is true for wood. This may be because the quick set u p time of a kerosene stove is appealing to those who need to leave for work quickly. In both villages , c har coal was the most popular fuel choice across all meals ( Figure 5 , Figure 6 ) . Electricity was used for a similar proportion of heating events across all meals once it had been introduced in phase 2 ( Figure 7 and Figure 8 ) . In both locations, breakfasts accounted for a smaller proportion of energy consumption than lunches and dinners ( Figure 9 , Figure 10 ). The median per capita energy consumptions for each type of heating event illustrate differences in the overall conversion efficiencies associated with different fuels ( Figure 11 and Figure 12 ). T

14 hese figures simply illustrate the valu
hese figures simply illustrate the values in Table 14 to Table 17 . F or most fuels, lunch and dinner were more energy intensive than breakfast on a per capita basis . Note that exceptions to this tended to be fuels that had not been used often for the meal in question. 11 A2EI Mini - grid Cooking Diaries Data Analysis These figures highlight an interesting feature – that per capita energy consumption of wood is highest in in Bezi, but in Mavota it is charcoal that has the highest per capit a consumption. This suggests that wood is used less efficiently in Bezi . Given that w ood is m ore commonly used in Bezi ( in Phase 1) , it may be more abundantly available, meaning cooks are less concerned about minimising its consumption, but this would need further investigation. Figure 5 Choice of fuels for heating events i n Bezi Figure 6 Choice of fuels for heating events i n Mavota 12 A2EI Mini - grid Cooking Diaries Data Analysis Figure 7 Choice of fuels for heating events per meal in Bezi Figure 8 Choice of fuels for heating events per meal in Mavota 13 A2EI Mini - grid Cooking Diaries Data Analysis Figure 9 Proportion of total energy consumption by heating event In Bezi Figure 10 Proportion of total energy consumption by heating event In Mavota Table 14 Per capita energy consumption by single fuel heating event s in Bezi – Phase 1 N Mean Median Std. Deviation 25% Quartile 75% Quartile Kerosene Breakfast 29 3.0 0.8 5.1 0.2 3.7 Lunch 10 5.0 1.5 7.2 0.4 9.3 Dinner 20 2.7 1.0 6.7 0.3 2.5 Charcoal Breakfast 127

15 4.6 3.8 3.6 2.3 6.0 Lunch
4.6 3.8 3.6 2.3 6.0 Lunch 158 4.7 4.3 3.4 2.5 6.1 Dinner 212 4.8 4.1 4.9 2.5 5.9 Wood Breakfast 5 7.8 8.6 4.1 4.0 11.2 Lunch 22 5.0 4.1 3.3 3.0 7.2 Dinner 24 10.2 6.5 11.7 3.5 10.4 14 A2EI Mini - grid Cooking Diaries Data Analysis Table 15 Per capita energy consumption by single fuel heating event s in Bezi – Phase 2 N Mean Median Std. Deviation 25% Quartile 75% Quartile Electricity Breakfast 5 0.2 0.2 0.1 0.2 0.3 Lunch 11 0.4 0.2 0.5 0.0 0.6 Dinner 6 0.3 0.3 0.3 0.1 0.6 Kerosene Breakfast 16 6.8 0.3 13.0 0.1 7.4 Lunch 4 3.8 1.2 6.0 0.3 9.9 Dinner 7 1.2 0.1 2.8 0.0 0.7 Charcoal Breakfast 127 5.6 3.3 8.4 1.6 6.0 Lunch 243 6.4 5.1 5.3 2.6 8.6 Dinner 278 6.4 5.0 7.3 2.8 7.8 Table 16 Per capita energy consumption by single fuel heating event s in Mavota – Phase 1 N Mean Median Std. Deviation 25% Quartile 75% Quartile Charcoal Breakfast 134 4.5 3.4 3.8 2.4 5.7 Lunch 339 6.3 5.0 4.6 3.0 7.8 Dinner 358 6.0 5.1 3.8 3.2 7.6 Wood Breakfast 11 3.0 2.9 1.4 1.9 4.2 Lunch 20 4.1 4.2 2.3 2.9 4.6 Dinner 16 3.5 2.9 2.1 1.8 4.3 Table 17 Per capita energy consumption by single fuel heating event s in Mavota – Phase 2 N Mean Median Std. Deviation 25% Quartile 75% Quartile Electricity Breakfast 4 0.1 0.1 0.1 0.1 0.2 Lunch 2 0.1 0.1 0.1 Dinner 5 0.2 0.2 0.1 0.1 0.2 Charcoal Breakfa

16 st 78 4.1 3.2 2.8 2.0 5.8
st 78 4.1 3.2 2.8 2.0 5.8 Lunch 325 5.5 4.8 3.8 2.8 7.3 Dinner 303 5.4 4.7 4.0 2.9 7.3 Wood Breakfast 15 2.6 2.3 1.4 1.7 2.9 Lunch 23 3.3 2.9 1.6 2.2 4.1 Dinner 12 3.1 2.9 1.6 2.2 4.0 15 A2EI Mini - grid Cooking Diaries Data Analysis Figure 11 Per capita median energy consumptions for different heating events in Bezi Figure 12 Per capita median energy consumptions for different heating events in Mavota 2.4 Energy consumption per day The to tal energy consumed per day has been calculated as the sum of the energy consumption of all heating events on a given date. T he number of household heating events is shown in Table 18 . The distributions of event energy consumption in Phase 1 and Phase 2 are presented in Figure 13 to Figure 16 . Note that all four of these charts only consider heating event s that used a single fuel , making the kerosene and electricity sample sizes quite small . 16 A2EI Mini - grid Cooking Diaries Data Analysis Table 18 Number of household - events from each town and phase Household events Bezi Phase 1 664 Bezi Phase 2 735 Mavota Phase 1 890 Mavota Phase 2 807 Total 3096 Figure 13 Distribution of daily energy consumption from wood in phase 1 in Bezi and Mavota (MJ/ event ) ( single fuels) Figure 14 Distribution of daily energy consumption from charcoal in phase 1 in Bezi and Mavota (MJ/ event ) ( single fuels) 17 A2EI Mini - grid Cooking Diaries Data Analysis Figure 15 Distribution of daily energy consumption from kerosene in phase 1 in Bezi and

17 Mavota (MJ/event) (single fuels) Fig
Mavota (MJ/event) (single fuels) Figure 16 Distribution of daily energy consumption from electricity in phase 2 in Bezi and Mavota (MJ/event) (single fuels) In Bezi, the total number of heating events in each phase did not v a ry much across the week ( Figure 17 ). There was slightly more fluctuation in wood and kerosene usage than charcoal, but this is most likely due to the small number of records using either wood or kerosene rather than a day - based trend. In Mav ota , the total number of heating events dip o n Friday and Saturday , and this pattern can be seen in both Phase 1 and Phase 2 ( Figure 18 ). Interestingly, this dip was no t evident in per capita energy consumption figures ( Figure 20 ), indicating that fewer people are cooked for on Friday and Saturday. Again, no trend for Bezi is evident from Figure 19 . The contribution of an EPC to the total daily energy consumption and total median daily energy consumption is extremely small. This shows th e potential for EPC s to be an extremely efficient cooking device. 18 A2EI Mini - grid Cooking Diaries Data Analysis Figure 17 Number of times each fuel was used throughout the week in Bezi (all households) Figure 18 Number of times each fuel was used throughout the week in Mavota (all households) 19 A2EI Mini - grid Cooking Diaries Data Analysis Figure 19 Median energy consumption per day for each fuel in Bezi Figure 20 Median energy consumption per day for each fuel in Mavota Daily household energy consumptions, based on these daily patterns of heating events, are presented in Table 19 . Note that these values have not been normalised for th e number o

18 f people catered for – rather, they
f people catered for – rather, they show the range of total daily energy consumptions at the household level, which is dependent on the number of people that each meal was prepared for. The mean number of household members catered for has been ca lculated for each day (across however many meals were prepared), and the table presents the mean of these means. This indicates that household wood and charcoal energy consumption s are similar, although wood is used to cook to feed more people. Although th e energy consumption when cooking with electricity is much lower, the number of records available is too small to draw any firm conclusions . 20 A2EI Mini - grid Cooking Diaries Data Analysis Table 19 Total daily energy consumption (MJ/household/day) – use of single fuel in a day Daily energy consumption (MJ/household/day) Proportion of days with heating event Household members (mean of means) n Mean Median Breakfast Lunch Dinner Wood (Phase 1) 38 68.4 55.3 36.8 84.2 76.3 8.3 Charcoal (Phase 1) 590 59.0 51.8 41.1 78.9 86.2 5.6 Kerosene (Phase 1) 20 32.8 16.3 68.4 36.8 68.4 5.4 Electricity (Phase 2) 10 0.7 0.6 20.0 20.0 60.0 6.3 2.5 EPC use during phase 2 On certain days in Figure 21 and Figure 22 , it appears as though EPCs are used for breakfast , lunch, and dinner. However as in Figure 23 and Figure 24 EPC usage is seldomly more than once a day, and so like kerosene stove, these 3 peaks are likely due to the profile being averaged over a month. There are two main characteristics of Figure 21 and Figure 22 : long lasting (�1hr), often high energy profiles and short er ( 1hr) bur

19 sts. It is likely that this is an indica
sts. It is likely that this is an indicator of the types of food participants are choosing to cook on their EPCs . I t is interesting to see that in Bezi, EPC use is mainly contained within the first two weeks of phase 2 ( Figure 23 ), whereas Figure 24 shows that in Mavota, there is a more consistent use of the EPCs throughout the month. Figure 21 24 - Hour Electric pressure cooker load profile averaged over each week in Bezi 21 A2EI Mini - grid Cooking Diaries Data Analysis Figure 22 24 - Hour Electric pressure cooker load profile averaged over each week in Mavota Figure 23 Number of EPC use s per day in Bezi over phase 2 Figure 24 Number of EPC use s per day in Mavota over phase 2 22 A2EI Mini - grid Cooking Diaries Data Analysis 3 Meals cooked 3.1 Food types cooked Separating out foods cooked for breakfast, lunch, or dinner only, Table 20 shows a high degree of consistency in the mix of foods cooked in both phases i.e., cooking with electricity did not cause participants to change what they cooked. There are a couple of exceptions e.g., in phase 2 , participants were more likely to cook rice or sweet potatoes/ c assava/ t aro root and in Bezi, fish stew. Table 21 shows which meals the most common dishes are cooked for. In Mavota, porridge is the most popular breakfast dish whereas in Bezi, households tend to cook either porridge or rice. Popular dishes tend to be the same across lunch and dinner with u gali being the most popular in both villages and phases. Otherwise, fish stew and dried dagaa are common Bezi dishe s and in Mavota, beans, leafy veg and dried dagaa are the next most popular. Interestingly

20 , in Mavota in phase 2, there is an incr
, in Mavota in phase 2, there is an increase in the amount of rice being cooked across all meals and in both Bezi and Mavota, there is an increase in s weet potatoes/ c assava/ t aro root being cooked at breakfast. Table 23 separates meals out by fuel used, and shows that EPCs were used to prepare rife and ugali for breakfasts and lunches, whereas bean were prepared almost exclusively as part of dinners. Table 22 shows that in Bezi in phase 2, participants were significantly less likely to prepare complex meals comprising of multiple dishes (Person chi square p 0.001) . If we assume this is because o f the introduction of an EtC, this could be explained by how easily ‘one pot’ meals can be prepared and cooked in a pressure cooker. It can be seen from Table 24 that some foods tend to be eaten on their own, such as bananas, chapati or fried fish . Others, such as beans , rice, and beef/ goat, tend to be eaten in more complex meals . 23 A2EI Mini - grid Cooking Diaries Data Analysis Table 20 Number of meals containing food type s (Breakfast, lunch , and dinner heating events only) Bezi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 Bananas 23 1.2% 16 0.8% 5 0.2% 2 0.1% 46 0.5% Beans 40 2.2% 36 1.8% 199 8.0% 181 8.2% 456 5.3% Beef/ goat 42 2.3% 64 3.1% 17 0.7% 37 1.7% 160 1.9% Beef/ goat grilled 4 0.2% 3 0.1% 7 0.3% 0 0.0% 14 0.2% Boiled potatoes 10 0.5% 1 0.0% 1 0.0% 12 0.5% 24 0.3% Cassava leaves 1 0.1% 4 0.2% 5 0.2% 0 0.0% 10 0.1% Chapati 9 0.5% 4 0.2% 6 0.2% 4 0.2% 23 0.3% Dagaa dried 91 4.9% 35 1.7%

21 208 8.4% 197 8.9% 531 6.2%
208 8.4% 197 8.9% 531 6.2% Dagaa fresh 36 1.9% 51 2.5% 2 0.1% 1 0.0% 90 1.0% Duck/ chicken stew 2 0.1% 5 0.2% 3 0.1% 10 0.5% 20 0.2% Fish/ stew 222 12.0% 289 14.2% 69 2.8% 53 2.4% 633 7.4% Fried cassava 3 0.2% 1 0.0% 2 0.1% 6 0.3% 12 0.1% Fried fish 42 2.3% 5 0.2% 5 0.2% 1 0.0% 53 0.6% Fried potatoes 8 0.4% 4 0.2% 12 0.5% 0 0.0% 24 0.3% Grilled fish 10 0.5% 5 0.2% 3 0.1% 18 0.8% 36 0.4% Leafy veg 36 1.9% 48 2.4% 160 6.4% 120 5.4% 364 4.2% Maandazi 5 0.3% 1 0.0% 1 0.0% 7 0.3% 14 0.2% Makande 14 0.8% 17 0.8% 16 0.6% 20 0.9% 67 0.8% Matoke 3 0.2% 2 0.1% 3 0.1% 1 0.0% 9 0.1% Mlenda 7 0.4% 6 0.3% 13 0.5% 0 0.0% 26 0.3% Okra 9 0.5% 9 0.4% 18 0.7% 0 0.0% 36 0.4% Pasta 23 1.2% 15 0.7% 1 0.0% 39 1.8% 78 0.9% Porridge 67 3.6% 72 3.5% 92 3.7% 39 1.8% 270 3.1% Pumpkin 28 1.5% 19 0.9% 47 1.9% 0 0.0% 94 1.1% Rice 106 5.7% 142 7.0% 74 3.0% 115 5.2% 437 5.1% Soup 3 0.2% 4 0.2% 2 0.1% 9 0.4% 18 0.2% Sweet potatoes/ Cassava/ Taro root 18 1.0% 85 4.2% 70 2.8% 87 3.9% 260 3.0% Ugali 361 19.5% 387 19.0% 577 23.2% 490 22.1% 1815 21.1% Other 624 33.8% 705 34.6% 868 34.9% 771 34.7% 2968 34.6% Total 1847 100% 2035 100% 2486 100% 2220 100% 8182 100% 24 A2EI Mini - grid Cooking Diaries Data Analysis Table 21 Top 10 food types separated by meal , village and phase

22 Bezi Mavota Total Pha
Bezi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 Breakfast Beans 10 6.5% 5 3.0% 7 5.1% 2 2.4% 24 4.4% Beef/ goat 9 5.8% 5 3.0% 0 0.0% 0 0.0% 14 2.6% Dagaa dried 5 3.2% 1 0.6% 0 0.0% 0 0.0% 6 1.1% Dagaa fresh 5 3.2% 1 0.6% 0 0.0% 0 0.0% 6 1.1% Fish/ stew 20 12.9% 21 12.6% 1 0.7% 2 2.4% 44 8.1% Leafy veg 1 0.6% 2 1.2% 1 0.7% 1 1.2% 5 0.9% Porridge 53 34.2% 53 31.7% 89 65.4% 38 45.8% 233 43.1% Rice 39 25.2% 47 28.1% 13 9.6% 15 18.1% 114 21.1% Sweet potatoes/ Cassava/ Taro root 10 6.5% 31 18.6% 19 14.0% 21 25.3% 81 15.0% Ugali 3 1.9% 1 0.6% 6 4.4% 4 4.8% 14 2.6% Total 155 100.0% 167 100.0% 136 100.0% 83 100.0% 541 100.0% Lunch Beans 6 1.6% 3 0.7% 90 14.5% 85 13.9% 184 8.9% Beef/ goat 16 4.1% 27 5.9% 6 1.0% 18 3.0% 67 3.2% Dagaa dried 43 11.1% 18 4.0% 101 16.3% 97 15.9% 259 12.5% Dagaa fresh 17 4.4% 25 5.5% 0 0.0% 1 0.2% 43 2.1% Fish/ stew 86 22.2% 105 23.1% 20 3.2% 21 3.4% 232 11.2% Leafy veg 14 3.6% 13 2.9% 75 12.1% 53 8.7% 155 7.5% Porridge 8 2.1% 18 4.0% 3 0.5% 0 0.0% 29 1.4% Rice 17 4.4% 28 6.2% 4 0.6% 19 3.1% 68 3.3% Sweet potatoes/ Cassava/ Taro root 6 1.6% 47 10.3% 43 6.9% 63 10.3% 159 7.7% Ugali 174 45.0% 171 37.6% 278 44.8% 253 41.5% 876 42.3% Total 387 100.0% 455 100.0% 620 100.0% 610 100.0% 2072 100.0% Dinner Beans 24 5.0% 28 4.8%

23 102 14.3% 94 15.0% 248 10.3%
102 14.3% 94 15.0% 248 10.3% Beef/ goat 17 3.6% 32 5.5% 11 1.5% 19 3.0% 79 3.3% Dagaa dried 43 9.0% 16 2.7% 107 15.0% 100 15.9% 266 11.1% Dagaa fresh 14 2.9% 25 4.3% 2 0.3% 0 0.0% 41 1.7% Fish/ stew 116 24.3% 163 27.8% 48 6.7% 30 4.8% 357 14.9% Leafy veg 21 4.4% 33 5.6% 84 11.8% 66 10.5% 204 8.5% Porridge 6 1.3% 1 0.2% 0 0.0% 1 0.2% 8 0.3% Rice 50 10.5% 67 11.4% 57 8.0% 81 12.9% 255 10.6% Sweet potatoes/ Cassava/ Taro root 2 0.4% 7 1.2% 8 1.1% 3 0.5% 20 0.8% Ugali 184 38.6% 215 36.6% 293 41.2% 233 37.2% 925 38.5% Total 477 100.0% 587 100.0% 712 100.0% 627 100.0% 2403 100.0% 25 A2EI Mini - grid Cooking Diaries Data Analysis Table 22 Number of foods included in a heating event (Breakfast, lunch and dinner heating events only) Bezi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 1 146 23.2% 206 28.4% 189 21.6% 175 22.2% 716 23.7% 2 392 62.2% 476 65.6% 655 75.0% 590 74.9% 2113 70.0% 3 79 12.5% 42 5.8% 24 2.7% 23 2.9% 168 5.6% 4 12 1.9% 2 0.3% 5 0.6% 0 0.0% 19 0.6% 5 1 0.2% 0 0.0% 0 0.0% 0 0.0% 1 0.0% Total 630 100% 726 100% 873 100% 788 100% 3017 100% 26 A2EI Mini - grid Cooking Diaries Data Analysis Table 23 Top 10 food types separated by meal and fuel choice (Phases 1 and 2) Charcoal Electricity Breakfast Beans 21 4.5% 1 2.4% Beef/ goat 13 2.8% 0 0.0% Dagaa dried 4 0.9% 0

24 0.0% Dagaa fresh 6 1.3% 0 0.
0.0% Dagaa fresh 6 1.3% 0 0.0% Fish/ stew 36 7.7% 8 19.0% Leafy veg 4 0.9% 1 2.4% Porridge 204 43.9% 1 2.4% Rice 92 19.8% 22 52.4% Sweet potatoes/ Cassava/ Taro root 75 16.1% 0 0.0% Ugali 10 2.2% 9 21.4% Total 465 100.0% 42 100.0% Lunch Beans 163 8.7% 7 9.0% Beef/ goat 64 3.4% 3 3.8% Dagaa dried 241 12.8% 1 1.3% Dagaa fresh 39 2.1% 0 0.0% Fish/ stew 205 10.9% 2 2.6% Leafy veg 143 7.6% 1 1.3% Porridge 27 1.4% 2 2.6% Rice 64 3.4% 38 48.7% Sweet potatoes/ Cassava/ Taro root 148 7.9% 9 11.5% Ugali 786 41.8% 15 19.2% Total 1880 100.0% 78 100.0% Dinner Beans 222 10.0% 17 53.1% Beef/ goat 74 3.3% 2 6.3% Dagaa dried 254 11.5% 2 6.3% Dagaa fresh 39 1.8% 0 0.0% Fish/ stew 317 14.3% 6 18.8% Leafy veg 195 8.8% 5 15.6% Porridge 8 0.4% 0 0.0% Rice 232 10.5% 0 0.0% Sweet potatoes/ Cassava/ Taro root 18 0.8% 0 0.0% Ugali 858 38.7% 0 0.0% Total 2217 100.0% 32 100.0% 27 A2EI Mini - grid Cooking Diaries Data Analysis Table 24 Occurrence of foods in meals by number of foods in the meal (all heating events, Phase 1 and 2) Dish occurrence 1 2 3 4 5 Bananas 29 13 3 1 Beans 13 47 33 7 1 Beef/ goat 5 127 28 1 Beef/ goat grilled 1 5 1 Boiled potatoes 6 5 1 Cassava leaves 1 3 1 Chapati 18 5 Dagaa dried 1 485 39 7 1 Dagaa fresh 61 26 3 Duck/ chicken stew 8 2 Fish/ stew 15 541 65 13 Fried cassava 4 2

25 Fried fish 41 9 3 Fried pot
Fried fish 41 9 3 Fried potatoes 7 4 1 Grilled fish 13 4 1 Leafy veg 4 293 61 9 Maandazi 1 2 4 Makande 6 7 3 1 Matoke 6 3 Mlenda 11 2 Okra 15 3 Pasta 24 12 3 Porridge 174 78 26 3 1 Pumpkin 25 2 3 Rice 11 295 3 4 Soup 2 5 1 1 Sweet potatoes/ Cassava/ Taro root 175 79 8 1 Ugali 29 1642 134 18 1 Other 6 44 13 4 Total 865 4123 465 7 4 3.2 Reheating food For each food item prepared, participants were asked if the dish was fresh, reheated, or partially cooked. Results for all those records that contained only a single heating event are presented in Table 25 . This shows that most meals are cooked fresh, and that breakfast s are most likely to be reheated , and that reheating foods appears to be more common practice in Bezi than in Mavota. Meals are rarely pa rtially cooked and saved for later. Preparing f ood for a baby is not included in th ese table s as there were only 3 records declaring this as the purpose, all of which were cooked from fresh. 28 A2EI Mini - grid Cooking Diaries Data Analysis Table 25 Number of meals fresh or rehea ted (single heating event records only) Totally Fresh Reheated Partially cooked Total Bezi Phase 1 Breakfast 112 74.7% 38 25.3% 150 100% Lunch 159 77.6% 46 22.4% 205 100% Dinner 213 79.2% 55 20.4% 1 0.4% 269 100% Total 486 77.4% 140 22.3% 2 0.3% 628 100% Phase 2 Breakfast 113 76.4% 35 23.6% 148 100% Lunch 233 87.9% 32 12.1% 265 100% Dinner 259

26 82.7% 54 17.3% 313 100% To
82.7% 54 17.3% 313 100% Total 606 83.4% 121 16.6% 727 100% Mavota Phase 1 Breakfast 122 93.8% 8 6.2% 130 100% Lunch 347 95.9% 14 3.9% 1 0.3% 362 100% Dinner 355 94.4% 21 5.6% 376 100% Total 824 94.9% 43 5.0% 1 0.1% 868 100% Phase 2 Breakfast 74 88.1% 10 11.9% 84 100% Lunch 347 96.1% 13 3.6% 1 0.3% 361 100% Dinner 317 94.3% 19 5.7% 336 100% Total 738 94.5% 42 5.4% 1 0.1% 781 100% Overall Total 5305 88.4% 691 11.5% 7 0.1% 6003 100% To find out which foods were most commonly reheated, cases in which a food was cooked in a meal were tagged as either fresh or reheated . Results collated across every dish are presented in Table 26 . In Bezi, beans, beef/ goat, f ried cassava , and soup were the dishes with the highest proport ion of reheats to freshly cooked , whereas in Mavota, it is hard to distinguish which dishes were commonly reheated compared to freshly cooked because of how rare r eheat ing was . In absolute terms, the foods most often reheated were ugali, and fish stew in Bezi, and beans in Mavota. 29 A2EI Mini - grid Cooking Diaries Data Analysis Table 26 Food types most commonly reheated (individual dishes, cooked as part of meals) Bezi Mavota Totally Fresh Reheated Totally Fresh Reheated Bananas 32 82.1% 7 17.9% 5 71.4% 2 28.6% Beans 43 55.1% 35 44.9% 341 88.3% 45 11.7% Beef/ goat 67 63.2% 39 36.8% 45 84.9% 8 15.1% Beef/ goat grilled 3 42.9% 4 57.1% Boiled potatoes 9 81.8% 2 18.2% 1 100.0% 0

27 0.0% Cassava leaves 4 80.
0.0% Cassava leaves 4 80.0% 1 20.0% Chapati 14 93.3% 1 6.7% 10 100.0% 0 0.0% Dagaa dried 114 89.1% 14 10.9% 398 98.3% 7 1.7% Dagaa fresh 72 82.8% 15 17.2% 3 100.0% 0 0.0% Duck/ chicken stew 6 85.7% 1 14.3% 2 66.7% 1 33.3% Fish/ stew 374 73.0% 138 27.0% 107 88.4% 14 11.6% Fried cassava 4 50.0% 4 50.0% 2 50.0% 2 50.0% Fried fish 42 89.4% 5 10.6% 6 100.0% 0 0.0% Fried potatoes 9 75.0% 3 25.0% Grilled fish 12 80.0% 3 20.0% 2 100.0% 0 0.0% Leafy veg 73 83.9% 14 16.1% 274 97.2% 8 2.8% Maandazi 3 60.0% 2 40.0% 2 100.0% 0 0.0% Makande 30 90.9% 3 9.1% 34 89.5% 4 10.5% Matoke 5 83.3% 1 16.7% 4 100.0% 0 0.0% Mlenda 12 92.3% 1 7.7% Okra 17 94.4% 1 5.6% Pasta 34 89.5% 4 10.5% 1 100.0% 0 0.0% Porridge 126 85.7% 21 14.3% 137 100.0% 0 0.0% Pumpkin 46 95.8% 2 4.2% Rice 172 68.5% 79 31.5% 166 86.9% 25 13.1% Soup 3 42.9% 4 57.1% 2 100.0% 0 0.0% Sweet potatoes/ Cassava/ Taro root 99 96.1% 4 3.9% 157 98.1% 3 1.9% Ugali 614 81.5% 139 18.5% 1017 95.0% 54 5.0% Other 19 79.2% 5 20.8% 42 97.7% 1 2.3% Total 1979 78.3% 547 21.7% 2837 94.1% 179 5.9% Reheating food for a meal might be expected to take less energy than preparing a meal from scratch (fresh), and indeed this can be seen to be the case for lunches and dinners for meals prepared using only charcoal in Phase 2 ( Table 27 ) , but not for breakfast. It is likely that fr

28 esh breakfast dishes do not require much
esh breakfast dishes do not require much heating and so reheating the previous night’s dinner may well consume more energy. Th is table show s that re heating lunches and dinners uses roughly two thirds of the energy taken to cook a fresh meal, but reheating breakfasts require one and half times the energy . 30 A2EI Mini - grid Cooking Diaries Data Analysis Table 27 Per capita energy consumption by heating event and reheating (MJ/pers/event) – Phase 2 Charcoal only N Mean Median Std. Deviation 25% Quartile 75% Quartile Breakfast Totally Fresh 155 4.9 3.3 5.7 1.8 6.0 Reheated 32 6.4 4.1 11.9 1.4 6.8 Lunch Totally Fresh 528 6.1 5.2 4.5 2.9 7.7 Reheated 39 3.5 2.4 4.6 1.3 4.3 Dinner Totally Fresh 513 6.1 5.0 6.1 3.0 7.7 Reheated 68 4.1 3.5 2.9 1.7 5.5 Food can only be reheated if it has been cooked in advance. Participants were asked to what extent food prepared in each heating event was then available for consumption at a later time ( Table 28 ) . Interestingly, while 11.5% of meals were claimed to be reheated, almost every meal was marked as have ‘none saved for later’. Lt may have been that participants misunderstood the options be low and perhaps a ‘there was some left over’ would have prompted a more consistent response. Table 28 Number of heating events that produce food that can be eaten later Save none for later Cooking in advance Precooking Total Bezi Phase 1 Breakfast 147 98.7% 1 0.7% 1 0.7% 149 100% Lunch 204 99.5% 1 0.5% 205 100% Dinner 267 100.0% 267 100% Bezi

29 Phase 2 Breakfast 148 100.0%
Phase 2 Breakfast 148 100.0% 148 100% Lunch 265 100.0% 265 100% Dinner 313 100.0% 313 100% Mavota Phase 1 Breakfast 129 100.0% 129 100% Lunch 355 98.3% 4 1.1% 2 0.6% 361 100% Dinner 367 97.6% 6 1.6% 3 0.8% 376 100% Mavota Phase 2 Breakfast 83 100.0% 83 100% Lunch 358 99.4% 1 0.3% 1 0.3% 360 100% Dinner 334 99.7% 1 0.3% 335 100% 31 A2EI Mini - grid Cooking Diaries Data Analysis 4 Cooking devices 4.1 Detail on how participants cook Participants were asked to record the following information on how they cooked: • Cooking device used i.e. what type of stove. • Type of cooking pot / utensil. • If a lid was used while cooking a dish . The cooking appliances used to cook individual foods are presented in Table 29 . Note that any single record (or meal) can contain information on up to nine foods, so the table includes each separate food – appliance combination. This shows that in both Bezi and Mavota, only 4 % of foods were cooked with electricity in Phase 2 . The choice of cooking utensils ( Table 30 ) and the use of a lid ( Table 31 ) apply only to heating events not using an EPC, so these would not be expected to change between phase 1 and phase 2 and, indeed, this appears to be the case. Table 29 Appliances used to cook foods (frequencies) Bezi Phase 1 Bezi Phase 2 Mavota Phase 1 Mavota Phase 2 Total Charcoal Stove 534 81.8% 679 91.5% 829 94.3% 734 89.4% 2776 89.7% EPC 31 4.2% 36 4.4% 67 2.2% Firewood Stove 68 10.4% 6 0.8% 50 5.7% 51

30 6.2% 175 5.7% Kero sene Stove
6.2% 175 5.7% Kero sene Stove 51 7.8% 26 3.5% 77 2.5% Total 653 100% 742 100% 879 100% 821 100% 3095 100% Table 30 Utensils used to cook foods (frequencies) Bezi Phase 1 Bezi Phase 2 Mavota Phase 1 Mavota Phase 2 Total Sufuria big 62 4.0% 53 3.3% 303 13.0% 300 14.4% 718 9.5% Sufuria medium 586 37.9% 658 40.8% 780 33.6% 704 33.7% 2728 36.0% Sufuria small 223 14.4% 172 10.7% 365 15.7% 305 14.6% 1065 14.1% Frying pan 35 2.3% 24 1.5% 2 0.1% 0.0% 61 0.8% Other 640 41.4% 706 43.8% 874 37.6% 780 37.3% 3000 39.6% Total 1546 100% 1613 100% 2324 100% 2089 100% 7572 100% Table 31 Use of lid when cooking foods (frequencies) Bezi Phase 1 Bezi Phase 2 Mavota Phase 1 Mavota Phase 2 Total Lid never used 7 1.1% 24 3.3% 21 2.4% 19 2.4% 71 2.3% Lid used in some dishes 42 6.6% 30 4.1% 12 1.4% 38 4.8% 122 4.0% Lid used in every dish 591 92.3% 673 92.6% 841 96.2% 734 92.8% 2839 93.6% Total 640 100% 727 100% 874 100% 791 100% 3032 100% 32 A2EI Mini - grid Cooking Diaries Data Analysis 4.2 Characteristics of different cooking devices It was noted earlier that there were two different load profiles to EPC use. A long, high powered profile, and a shorter smaller on e (see Section 2.5 ) . Table 32 considers EPC usage in both Bezi and Mavota and helps to explain that the longer EPC usage is most likely due to the cooking of beans and the shorter represents cooking rice. It is interestin g that there is such a clear difference between

31 which dishes participants cho o se
which dishes participants cho o se to cook in an EPC and which dishes they cooked on other stoves. If the rice they were cooking was whole grain they may well have noticed the time benefit of EPCs when cooking whole grain rice and/or beans. T he lack of diversity in the dishes cooked using an EPC is most likely due to a lack of awareness about what can be cooked with an EPC. Table 32 Cooking devices used to cook different food types (frequencies) Charcoal Wood Kerosene EPC Phase 1 Phase 2 Phase 1 Phase 2 Phase 1 Phase 2 Phase 2 Bananas 25 13 1 2 5 Beans 226 196 17 19 2 25 Beef/ goat 56 98 3 1 2 4 5 Beef/ goat grilled 3 3 1 Boiled potatoes 9 1 2 Cassava leaves 1 3 1 Chapati 14 6 3 1 1 Dagaa dried 279 226 25 7 2 3 Dagaa fresh 35 52 1 2 Duck/ chicken stew 1 8 1 Fish/ stew 244 329 39 6 20 7 7 Fried cassava 4 1 1 Fried fish 44 6 3 1 Fried potatoes 6 4 2 Grilled fish 14 5 Leafy veg 192 159 13 6 2 2 7 Maandazi 6 1 Makande 29 38 2 1 1 1 Matoke 7 2 1 Mlenda 7 6 Okra 8 5 2 5 Pasta 18 15 2 4 Porridge 151 108 15 9 2 1 1 Pumpkin 28 17 1 2 Rice 172 235 10 7 7 9 35 Soup 5 4 Sweet po tatoes/ Cassava/ Taro root 82 162 4 6 2 8 Ugali 851 8

32 40 88 35 23 9 15 N.B. intens
40 88 35 23 9 15 N.B. intensity of green shading reflects the proportion of foods cooked using a single fuel in a single phase. 33 A2EI Mini - grid Cooking Diaries Data Analysis 4.3 Fuel stacking The number of meal preparation records in which single and multiple cooking appliances were used are presented in Table 33 . In Mavota, there was a 3% increase in th e number of records using 2 appliances per meal in Phase 2 . Table 34 shows that 55% of records involving the use of an EPC were part of a stack , so the increase seen in appliance stacking is due to the introduction of EPCs into Mavota. Table 33 Number of occurrences cooking devices were used a specified number of times in preparing meals Number of cooking devices Bezi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 1 627 98.0% 712 97.9% 867 99.3% 761 96.2% 2967 97.9% 2 5 13 2.0% 15 2.1% 6 0.7% 30 3.8% 64 2.1% Total 640 100% 727 100% 873 100% 791 100% 3031 100% Table 34 Total number of times c ooking devices were used in preparing meal s Number of devices used Charcoal stove EPC Wood stove Kerosene stove Total Bezi Phase 1 1 521 98% 55 81% 51 100% 627 70% 2 13 2% 13 19% 26 5% Total 534 100% 68 100% 51 100% 653 100% Bezi Phase 2 1 665 98% 20 65% 3 50% 24 92% 712 76% 2 14 2% 11 35% 3 50% 2 8% 30 24% Total 679 100% 31 100% 6 100% 26 100% 742 100% Mavota Phase 1 1 823 99% 44 88% 867 47% 2 6 1% 6 12% 12 3% Total 829 100%

33 50 100% 879 100% Mavo
50 100% 879 100% Mavota Phase 2 1 706 96% 10 28% 45 88% 761 53% 2 28 4% 26 72% 6 12% 60 22% Total 734 100% 36 100% 51 100% 821 100% Total 1 2715 98% 30 45% 147 84% 75 97% 2967 81% 2 61 2% 37 55% 28 16% 2 3% 128 19% Total 2776 100% 67 100% 175 100% 77 100% 3095 100% 4.4 Energy used by different stoves Table 35 shows th e per capita energy consumption on each stove across both phases and villages . As seen earlier, electricity is the most efficient fuel used by participants, using roughly one 20 th of the energy used by a charcoal stove ( by median ) . The energy required per capita to cook various dishes on charcoal stoves and EPC s is seen in Table 36 . The table omit s the energy consumption of any dish that only occurred once. All three dishes that are 5 For multiple devices, Table 33 states how many times two cooking devices were used in a single record. Table 34 counts how often each type of cookstove was recorded. For rows in which two devices were used, records are counted twice – once for each type of stove. These row totals are, therefore, double the number of records given in Table 33 . 34 A2EI Mini - grid Cooking Diaries Data Analysis regularly cooked in EPC s use less than one 10 th of the energy per capita used by a charcoal stove , and both beans and rice use about one 30 th of the energy. Table 35 Per capita energy consumption (MJ/pers/event) of single meals cooked using different devices Cooking device Frequency Mean Median Std.dev. 25% Quartile 75% Quartile Charcoal stove 362

34 5.0 3.8 7.8 2.3 6.6 EPC 2
5.0 3.8 7.8 2.3 6.6 EPC 23 0.3 0.2 0.4 0.1 0.4 Firewood stove 33 4.1 2.9 6.5 2.0 3.7 Kerosene stove 10 1.5 0.2 2.6 0.1 2.1 Table 36 P er capita energy consumption of single dishes on different stoves (MJ/pers/event) Food(s) Frequency Mean Median Std.dev 25% Quartile 75% Quartile Charcoal Stove Bananas 7 7.0 6.3 5.5 2.1 14.1 Beans 2 12.9 12.9 1.4 11.9 Beef / g oat 2 4.0 4.0 2.4 2.3 Boiled / p otatoes 3 4.0 3.2 2.9 1.6 Chapati 4 15.1 13.4 13.8 2.9 29.0 Fish s tew 3 8.8 9.2 4.1 4.5 Makande 51 8.5 7.9 13.3 4.5 14.2 Other 5 5.5 4.4 6.5 0.7 11.0 Pasta 3 8.0 4.3 8.6 1.9 Porridge 116 3.6 3.0 6.0 1.9 4.5 Pumpkin 22 4.3 3.6 3.0 2.2 6.8 Rice 25 2.9 5.1 11.8 3.1 8.5 Sweet / p otatoes / c assava / t aro r oot 94 4.4 3.2 3.9 2.2 5.4 Ugali 16 5.6 4.7 4.5 1.7 9.9 Electric Pressure cooker Beans 8 0.43 0.31 0.43 0.10 0.70 Rice 5 0.10 0.12 0.06 0.05 0.15 Sweet / p otatoes / c assava / t aro r oot 7 0.48 0.29 0.50 0.17 0.61 5 Time taken 5.1 Time taken to cook food types The times taken to cook individual food types using only charcoal are presented in Table 37 . Note that this includes times taken to cook individual foods when prepared as part of a multi - dish meal. This suggests four groups of foods: • Quick , less than 30 mins – d agaa , grilled fish, veg, mandazi, mlenda, Matoke okra . • 30 mins to an hour – bananas, beef/ goat, rice, potatoes , chapati, duck/ chicken,

35 pasta, porridge. 35 A2EI
pasta, porridge. 35 A2EI Mini - grid Cooking Diaries Data Analysis • Longer , an hour or more – c assava leaves, beans , m akande , soup . When looking at the time spent cooking dishes in an EPC ( Table 38 ), it is clear that EPC s have the potential to shorten cooking times. The mean average cooking time for rice drops from about 40 minutes to 30; for beef/goat and Sweet potatoes/ taro root the time drop s from 40mins to 20mins; for beans, the drop is even larger from almost 2 and a half hours to less than an hour. On the other hand, fish stew appears to take a little longer in a pressure cooker, although it should be noted that there are only a few records for cooking fish stew in a pressure cooker. These figures highlight the importance of educating cooks on how to use an EPC most effectively. Although the EPC takes time to reach pressure, the cooking time at pressure is much reduced . Cooks may initially se t the EtC close to a ‘normal’ cooking time, but then it takes time and experience to work out an acceptable minimum cooking time. Table 37 Time taken to cook food types using c harcoal stove only ( hh:mm ) N Mean Median Std. Deviation 25% Quartile 75% Quartile Bananas (hard) 37 00:40 00:35 00:28 00:22 00:53 Beans 404 02:26 02:46 01:13 01:55 03:16 Beef/Goat 147 00:44 00:38 00:30 00:19 01:03 Boiled potatoes 10 00:46 00:37 00:26 00:31 01:01 Cassava leaves 4 02:09 01:58 01:06 01:14 03:17 Chapati 20 00:49 00:46 00:33 00:21 01:08 Dagaa (dried) 499 00:22 00:20 00:21 00:15 00:25 Dagaa (fresh) 85 00:29 00:29 00:19 00:15 00:40 Duck/chicken st

36 ew 9 00:46 00:31 00:41 00:15
ew 9 00:46 00:31 00:41 00:15 01:12 Fish stew (boiled) 564 00:33 00:28 00:33 00:15 00:45 Fried cassava 5 00:54 00:29 00:44 00:21 01:39 Fried fish 50 00:42 00:30 00:54 00:18 00:46 Fried potatoes 10 00:34 00:34 00:09 00:26 00:42 Grilled fish 6 00:21 00:27 00:13 00:05 00:33 Grilled goat/beef 19 00:51 00:40 00:43 00:16 01:23 Leafy veg 345 00:21 00:18 00:14 00:14 00:26 Maandazi 7 00:22 00:23 00:18 00:05 00:40 Makande 64 03:20 03:32 01:15 02:38 04:14 Matoke 9 00:29 00:26 00:16 00:17 00:42 Mlenda 13 00:25 00:20 00:11 00:15 00:37 Okra 11 00:22 00:25 00:09 00:15 00:29 Other 63 00:53 00:29 00:55 00:17 01:07 Pasta 32 00:34 00:29 00:18 00:24 00:41 Porridge 250 00:36 00:35 00:14 00:28 00:43 Pumpkin 44 00:55 00:53 00:25 00:43 01:08 Rice 383 00:41 00:35 00:31 00:25 00:48 Soup (goat, beef, fish) 9 01:00 00:35 00:48 00:21 01:51 Sweet potatoes/ taro root 236 00:46 00:43 00:22 00:32 00:56 Ugali 1680 00:32 00:30 00:18 00:22 00:39 36 A2EI Mini - grid Cooking Diaries Data Analysis Table 38 Time taken to cook food types using e lectric pressure cooker only (hh:mm) N Mean Median Std. Deviation 25% Quartile 75% Quartile Beans 18 00:55 00:48 00:47 00:29 01:02 Beef / g oat 4 00:22 00:23 00:09 00:12 00:31 Fish s tew 5 00:41 00:39 00:21 00:21 01:03 Rice 28 00:30 00:30 00:15 00:20 00:33 Sweet potatoes/ taro root 7 00:19 00:18 00:03 00:17 00:23 5.2 Time taken to prepare meal Inter

37 estingly , the shortened cooking time
estingly , the shortened cooking time seen when using EPC s only really effects the time taken to cook breakfast and lunch, not dinner (comparing Table 40 with Table 39 ) . Care needs to be taken in interpreting these differences, as it can be seen from Table 23 that there a re differences in the foods cooked using these two devices. For example, when cooking lunch e s using charcoal, ugali is the most common food, but when cooking with an EPC, rice is most commonly co oked (followed by ugali). When cooking dinners using charcoal , ugali is the most common food, but when cooking with an EPC, beans, which take a long time, are most commonly cooked. This appears to confirm that some cooks may be unaware of how to cook ugali in the EPC. Table 39 Duration of heating events using c harcoal stove only (hh:mm) N Mean Median Std.dev. 25% Quartile 75% Quartile Breakfast 550 00:37 00:32 00:27 00:20 00:44 Lunch 2029 00:47 00:30 00:53 00:20 00:46 Dinner 2375 00:44 00:30 00:47 00:20 00:44 Table 40 Duration of heating events using e lectric pressure cooker only (hh:mm) N Mean Median Std.dev. 25% Quartile 75% Quartile Breakfast 8 00:20 00:16 00:11 00:11 00:24 Lunch 24 00:29 00:24 00:19 00:18 00:33 Dinner 37 00:45 00:32 00:36 00:30 00:59 5.3 Time of day Households using an EPC usually began cooking breakfast almost an hour later than charcoal stove users and started cooking lunch and dinner an hour earlier (comparing Table 42 with Table 41 ) . The change in cooking time could be caused by a number of factors , e.g. when the quality of electricity supply is best ; the dishes cooked (as s

38 een earlier; the dishes cooked on EPC s
een earlier; the dishes cooked on EPC s are often different to other stoves); an acknowledgement of the shortened cooking time and thus an adjustment of the meal start time (for breakfasts), but this is speculation and requires further investigation . 37 A2EI Mini - grid Cooking Diaries Data Analysis Table 41 Time of day to start preparing meal using charcoal stove only (hh:mm) N Mean Median Std.dev. 25% Quartile 75% Quartile Breakfast 550 08:50 08:11 02:41 07:40 08:59 Lunch 2026 12:07 12:15 02:16 11:02 13:10 Dinner 2374 18:40 19:16 01:52 18:03 19:55 Table 42 Time of day to start preparing meal using electric pressure cooker only (hh:mm) N Mean Median Std.dev. 25% Quartile 75% Quartile Breakfast 8 10:10 08:54 03:55 08:29 09:26 Lunch 23 11:08 11:10 01:47 09:37 12:36 Dinner 36 17:51 18:15 02:04 16:26 19:27 6 Water heating There were no water heating events whose sole purpose was to heat water, instead records indicated that water was heated as part of preparing a breakfast, lunch, or dinner . Wa ter was most commonly heated for a beverage, sometimes for bathing or washing hands ( Table 43 ). A strange phenomenon occurs in Mavota during phase 2 where heating water for bathing seems to stop. 93% of records only contained 1 water heating event, the other 7% contained 2 with bathi ng being more common in records with 2 water heating events. Hot beverages were most often prepared at breakfast time, and it was most common for participants to heat water for bathing in the evening at dinner time ( Table 44 ). Table 43 Reason for heated water (all records) Be

39 zi Mavota Total Phase 1 Phase 2
zi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 Drinking/purifying 5 2% 0 0% 1 1% 0 0% 6 1% Tea/coffee/cocoa/milk 198 73% 168 89% 54 50% 52 93% 472 76% Washing hands 17 6% 12 6% 1 1% 0 0% 30 5% Bathing 35 13% 6 3% 49 46% 4 7% 94 15% Other 15 6% 3 2% 2 2% 0 0% 20 3% Total 270 1 00% 189 100% 107 100% 56 100% 622 100% 38 A2EI Mini - grid Cooking Diaries Data Analysis Table 44 Purpose of water heating by heating events (single heating events only) Drinking/ purifying Tea/ coffee/ cocoa/ milk Washing hands Bathing Other Total Bezi Phase 1 Breakfast 0 0% 130 76% 0 0% 1 4% 1 11% 30 65% Lunch 1 25% 10 6% 3 21% 2 8% 3 33% 3 7% Dinner 3 75% 31 18% 11 79% 23 88% 5 56% 13 28% Total 4 100% 171 100% 14 100% 26 100% 9 100% 46 100% Bezi Phase 2 Breakfast 126 76% 1 10% 1 17% 0 0% 31 78% Lunch 17 10% 2 20% 0 0% 1 50% 3 8% Dinner 22 13% 7 70% 5 83% 1 50% 6 15% Total 165 100% 10 100% 6 100% 2 100% 40 100% Mavota Phase 1 Breakfast 1 100% 43 83% 0 0% 3 6% 0 0% 9 69% Lunch 0 0% 4 8% 0 0% 2 4% 1 50% 1 8% Dinner 0 0% 5 10% 1 100% 42 89% 1 50% 3 23% Total 1 100% 52 100% 1 100% 47 100% 2 100% 13 100% Mavota Phase 2 Breakfast 45 90% 90% 0 11 73% Lunch 4 8% 8% 0 2 13% Dinner 1 2% 2% 4 2 13%

40 Total 50 100% 100% 4
Total 50 100% 100% 4 15 100% As found with cooking meals, charcoal stoves are most commonly used for heating water ( Table 45 ). Even though it is not what they are designed for, participants did make some use of EPCs to heat water in Phase 2. Table 46 shows that EPCs and kerosene stove s were almost exclusively used for beverages, where speed is important. Despite the small number of records, results indicates that EPC s are more efficient th an charcoal stoves , consuming less than one 20 th of the mean energy per capita required to make a beverage ( Table 47 ). Table 45 Devices used to heat water (single use of water in water heating events only) Bezi Mavota Total Phase 1 Phase 2 Phase 1 Phase 2 Charcoal stove 187 80% 162 89% 104 99% 51 91% 504 87% Firewood st ove 21 9% 0 0% 1 1% 0 0% 22 4% Kerosene stove 27 11% 13 7% 0 0% 0 0% 40 7% Pressure cooker 0 0% 8 4% 0 0% 5 9% 13 2% Total 235 1 00% 183 100% 105 100% 56 100% 579 100% 39 A2EI Mini - grid Cooking Diaries Data Analysis Table 46 Reason for heating water on different stoves Drinking/ purifying Tea/ coffee/ cocoa/ milk Washing hands Bathing Other Total Bezi Charcoal stove 3 309 20 37 11 380 60% 84% 69% 90% 61% 83% Firewood stove 2 9 9 3 2 25 40% 3% 31% 7% 11% 5% Kerosene stove 0 40 0 1 5 46 0% 11% 0% 2% 28% 10% Pressure cooker 0 8 0 0 0 8 0% 2% 0% 0% 0% 2% Total 5 366 29 41 18 459 100% 100% 100% 100% 100% 100% Mavot

41 a Charcoal stove 1 101 1 52
a Charcoal stove 1 101 1 52 2 157 100% 95% 100% 98% 100% 96% Firewood stove 0 0 0 1 0 1 0% 0% 0% 2% 0% 1% Pressure cooker 0 5 0 0 0 5 0% 5% 0% 0% 0% 3% Total 1 106 1 53 2 163 100% 100% 100% 100% 100% 100% Table 47 Per capita energy consumed by heating water for Tea/coffee/cocoa/milk (single use of only) (MJ/pers/event) Frequency Mean Median Std.dev. 25% Quartile 75% Quartile Charcoal stove 43 3.42 2.49 2.77 1.68 4.60 Electric pressure cooker 4 0.15 0.18 0.08 0.08 0.21 Table 48 and Table 49 show that non - electric k ettles were rarely used to heat water , but when they were, a l id was al ways used. Medium sufurias were the most popular utensil to heat water in and along with small sufurias, often used lids, large sufurias only used a lid 39% of the time and were usually used to heat bath water . It was not common for heated water to be stor ed in a flask to be save d for later ( Table 50 ) , but when it was, it was always for hot beverage s . Table 48 Utensils used to heat water (all bioma ss water heating events) Drinking/ purifying Tea/ coffee/ cocoa / milk Washing hands Bathing Other Total Non - electric Kettle 0 0% 7 2% 0 0% 0 0% 0 0% 7 1% Sufuria big 0 0% 3 1% 1 4% 44 52% 2 10% 48 8% Sufuria medium 4 80% 251 56% 10 40% 34 40% 14 70% 311 54% Sufuria small 1 20% 176 39% 14 56% 6 7% 4 20% 200 35% Total 5 100% 450 100% 25 100% 84 100% 20 100% 579 100% 40 A2EI Mini - grid Cooking Di

42 aries Data Analysis Table 49 Use
aries Data Analysis Table 49 Use of lids (all biomass water heating events) Kettle Sufuria big Sufuria medium Sufuria small Total No 0 0% 30 63% 33 11% 20 10% 83 14% Yes 7 100% 18 38% 278 89% 180 90% 496 86% Total 7 100% 48 100% 311 100% 200 100% 579 100% Table 50 Storing heated water in a flask (all water heating events) Stored in a flask All 3% 15 Some 2% 10 None 96% 554 Total 100% 579 Unsurprisingly, h eating water for washing hands takes the shortest amount of time and for bathing takes the longest ( Table 51 ). Kerosene stoves and EPC s we re both used mostly for heating beverages, and results in Table 52 indicates that kerosene stoves are quicker. Table 51 Time taken to heat wate r for different purposes (minutes) Frequency Mean Median Std.dev. 25% Quartile 75% Quartile Drinking/purifying 6 00:19 00:12 00:22 00:01 00:36 Tea/coffee/cocoa/milk 454 00:22 00:19 00:14 00:13 00:29 Washing hands 29 00:10 00:05 00:13 00:03 00:12 Bathing 91 00:26 00:21 00:17 00:15 00:34 Other 19 00:22 00:10 00:24 00:04 00:41 Table 52 Time taken to heat water using different devices (minutes) Frequency Mean Median Std.dev. 25% Quartile 75% Quartile Charcoal stove 517 00:23 00:20 00:15 00:13 00:30 Firewood stove 26 00:15 00:09 00:15 00:01 00:30 Kerosene stove 44 00:18 00:14 00:16 00:09 00:21 Pressure cooker 12 00:23 00:17 00:17 00:15 00:24 7 Conclusion From the cooking diaries carried out by the participant in Bezi and Mavota over two phases, 3 features of cookin

43 g with an EPC were made evident: â
g with an EPC were made evident: • EPC efficiency • Choice of foods cooked in an EPC • Shortened length of time cooking when using an EPC 41 A2EI Mini - grid Cooking Diaries Data Analysis Throughout this report, various tests have been made to check the energy consumption of the different fuels and stoves, be it across meals, weekdays, dishes etc. In every test the EPC was found to use roug hly 1/ 2 0 th of the energy used by charcoal stoves, the most popular stove. Table 32 showed that the proportion of each dish cooked in Phase 1 and Phase 2 stayed the sa me across the traditional stoves. However, when it came to EPC s, there was a shift from popular dishes like ugali to beans and rice. Further investigation is required to understand w hy this shift happened ; it may be, for example, that rice and beans were u sed as example dishes when the participants were given the pressure cookers, or maybe people were already familiar with rice cookers , which made the transition to cooking rice in a pressure cooker easy. The length of time needed to cook certain dishes was shortened a lot when using an EPC , most notably when cooking beans , which took less than half the time. This is potentially an extremely attractive quality of cooking with an EPC , that time can be saved and spent on other tasks. However as mentioned at th e start of this report, despite the benefits of EPC s, ultimately cost will dictate behaviour. These pressure cookers were rarely used, and this was largely due to the high mini - grid tariffs at the time of this experiment. Until the cost of electricity delivered by mini grids can compete with traditional fuels, it is unlikely that