0.1 About the report

This is a monthly report bringing in all the basic data in a readable format, using charts and tables. We have used statistical tools to handle missing data and added our own conclusions as summary ahead of each table or chart. There are following sections in the entire report.

We will add more sections after brainstorming with Power plant SMEs and Airtel Management.

The present report has a total of 151 days data in the range Oct 1st to Feb 28th for a total of 5 sites. The input to this report was raw data of fuel levels, events of DG, AC and DC electrical parameters, all polled at 1 minute frequency.

0.2 AC Source Energy

0.2.1 Grid Energy

To start with let’s take a look at total AC kWh clocked for each site for each month.

Monthly GRID Energy used (kWh)
Oct'21 Nov'21 Dec'21 Jan'22 Feb'22
BagruUPG1 8,003 7,030 6,986 6,725 5,880
BagruUPG2 5,114 4,399 4,266 4,109 4,018
Dand01 5,005 5,193 5,249 3,506 3,333
Jobner 7,489 7,056 7,085 6,813 6,507
Sawarda 3,045 2,638 3,340 3,225 2,927
TOTAL 28,657 26,316 26,925 24,377 22,664

0.2.2 Normalised grid energy per month

Due to variable missing data each month and number of days changing every month, the grid kwh data may not give a representative trend. Therefore a normalized table with exact 30 days kWh is computed by apportioning the consumption to exactly 720 hours in each month.

0.2.3 Trend of grid energy.

If we plot all the EB Watt hours logged each day in a box plot, the trend of WH variations can be visualized much better in compact boxplots. The below boxplot shows that data consistency is good for all sites except in Dand where in October it wildly fluctuated. This will need more investigation. Bagru UPG1 is one site that ahows a clear downward trend in EB consumption and other sites seem to be closer to stationary in their seasonality.

Note on boxplot defaults. The taller the box the greater is the volatility of the data points (kWh each day). The median line is the solid horizontal line inside each box. The lower and upper hinges correspond to the first and third quartiles (the 25th and 75th percentile thresholds). The upper whisker extends from the hinge to the largest value no further than 1.5 times IQR from the hinge (where IQR is the inter-quartile range, or distance between the first and third quartiles). The lower whisker extends from the hinge to the smallest value at most 1.5 times of IQR of the hinge. Data beyond the end of the whiskers are called “outlying” points and are plotted individually.

0.2.4 DG energy

Following is the kWh generated month wise by DG sets on each site. Note: when grid supply is off either battery or DG supplies energy. Here we are focussing on DG supplied energy.

Monthly DG Energy used (kWh)
Oct'21 Nov'21 Dec'21 Jan'22 Feb'22
BagruUPG1 1,097 1,067 998 1,024 963
BagruUPG2 774 759 748 713 726
Dand01 904 519 696 585 239
Jobner 922 701 607 809 559
Sawarda 994 654 667 814 791
TOTAL 4,691 3,701 3,716 3,945 3,277

0.2.4.1 DG Energy as percentage of total

This is the split between Grid and DG energy aggregated over the entire 5 months time span.

DG Energy Generation as percent of total
Across the months Oct'21 to Feb'22}
Grid DG Percent
DG
BagruUPG1 34,624 5,149 12.95%
BagruUPG2 21,906 3,720 14.50%
Dand01 23,288 2,943 11.20%
Jobner 34,949 3,597 9.35%
Sawarda 15,174 3,920 20.55%

0.3 Run hours

This is the summary of run hours (usage time) of various energy sources on a monthly basis.

Run hours across each energy source
Month Grid(EB) DG BAT
Sawarda
Sawarda Oct'21 546.47 168.87 23.09
Sawarda Nov'21 490.25 139.85 16.20
Sawarda Dec'21 588.50 115.74 36.14
Sawarda Jan'22 571.98 137.46 30.03
Sawarda Feb'22 510.06 126.46 30.92
Average — 541.45 137.68 27.28
Jobner
Jobner Oct'21 662.11 77.41 4.00
Jobner Nov'21 653.68 63.42 1.52
Jobner Dec'21 679.55 62.11 5.00
Jobner Jan'22 664.13 75.02 2.53
Jobner Feb'22 607.58 49.55 11.53
Average — 653.41 65.50 4.92
Dand
Dand01 Oct'21 597.63 88.62 57.75
Dand01 Nov'21 628.22 51.39 40.39
Dand01 Dec'21 616.63 71.03 56.34
Dand01 Jan'22 629.24 64.52 50.24
Dand01 Feb'22 620.29 31.76 19.95
Average — 618.40 61.46 44.93
Bagru2
BagruUPG2 Oct'21 620.10 108.28 1.77
BagruUPG2 Nov'21 622.04 102.77 0.10
BagruUPG2 Dec'21 639.36 112.19 0.41
BagruUPG2 Jan'22 637.28 112.07 0.22
BagruUPG2 Feb'22 576.05 101.55 1.26
Average — 618.97 107.37 0.75
Bagru1
BagruUPG1 Oct'21 645.99 92.41 3.19
BagruUPG1 Nov'21 622.91 94.74 0.78
BagruUPG1 Dec'21 640.85 94.24 6.36
BagruUPG1 Jan'22 611.90 106.81 4.02
BagruUPG1 Feb'22 568.17 97.56 1.39
Average — 617.96 97.15 3.15

Note: the total hours in a month are 720 (744 for 31 days month). Some sites monthly totals add up to slightly higher than the gross hours in the month due to overlapping/ simultaneous run times.

0.3.0.1 Comparison of run hours

Run hour trends can be seen in this plot for all 3 energy sources. As you see maximum Grid energy is used in Jobner, max DG energy in Sawarda and max Battery is deployed in Dand. Battery is hardly used in Bagru and Jobner sites.

0.4 Box Plot of DG run hours

Run hour box plots reveal a lot more information about the distribution of DG run hours. As we see from the 5 boxplot trends, Dand & Jobner have the lowest medians. Dand has a very close lower hinge in the box which means most data is close to the lowest data. We also see many outliers of high values in Dand. Since Dand has such a low median and 25 percentile line, it means the DG is getting fired several times for very short duration, which could be inefficient not just for DG battery life but for fuel economy also. Sawarda DG run hours are the most predictable with few outliers.

0.5 Diesel Generator Metrics

We have summarised the total consumption, total refuel (net of drain), total run hours, idling hours and finally CPH. CPH has been calculated by two methods.

The Second method will be closer to the actual CPH, as it excludes the idling run hours.

site_name Month Cons.
Actual(L)		 Refuel
gross(L)		 Refuel
Net IN(L)		 Drain
gross(L)		 DG Run hrs
Ignition		 DG Run hrs
Connected		 DG Idle hrs
Detected1		 DG Load
Average kWatts		 CPH
Net (Fuel/Loadhrs) (L/Hr)		 Eff. Ratio
EER (WH/L)
site_name
BagruUPG1 Oct'21 389.8 419.5 401.7 17.8 92.41 83.61 3.99 243.9 4.90 2,808
BagruUPG1 Nov'21 392.1 397.6 397.6 0.0 94.74 76.80 3.56 868.5 5.35 2,703
BagruUPG1 Dec'21 374.3 352.4 352.4 0.0 94.24 86.35 5.35 592.0 4.62 2,673
BagruUPG1 Jan'22 393.6 405.3 405.3 0.0 106.81 96.87 4.90 106.3 4.28 2,606
BagruUPG1 Feb'22 393.3 452.0 366.8 85.2 97.56 93.22 4.35 146.1 4.43 2,446
BagruUPG2 Oct'21 407.1 430.8 430.8 0.0 108.28 88.47 14.16 189.1 5.48 1,954
BagruUPG2 Nov'21 354.6 405.3 379.1 26.2 102.77 82.36 8.77 544.5 4.82 2,156
BagruUPG2 Dec'21 405.5 358.1 358.1 0.0 112.19 97.06 15.13 269.6 4.95 1,828
BagruUPG2 Jan'22 403.6 401.3 401.3 0.0 112.07 100.14 11.97 77.9 4.58 1,759
BagruUPG2 Feb'22 379.6 419.8 400.2 19.6 101.55 87.29 14.22 135.9 5.20 1,912
Dand01 Oct'21 360.4 376.7 274.6 102.1 88.62 0.00 0.00 NA Inf 2,720
Dand01 Nov'21 198.3 179.4 179.4 0.0 51.39 0.00 0.00 NA Inf 2,651
Dand01 Dec'21 261.3 395.8 395.8 0.0 71.03 0.00 0.00 NA Inf 2,715
Dand01 Jan'22 246.8 167.4 167.4 0.0 64.52 0.00 0.00 NA Inf 2,575
Dand01 Feb'22 112.7 202.6 202.6 0.0 31.76 0.00 0.00 NA Inf 2,446
Jobner Oct'21 336.3 318.4 318.4 0.0 77.41 63.10 4.09 907.2 5.70 2,756
Jobner Nov'21 262.5 325.4 325.4 0.0 63.42 57.25 6.17 650.7 5.14 2,668
Jobner Dec'21 153.4 352.5 352.5 0.0 62.11 49.81 10.18 190.6 6.35 2,879
Jobner Jan'22 298.7 240.5 240.5 0.0 75.02 66.21 6.76 221.9 5.02 2,730
Jobner Feb'22 212.8 278.9 210.7 68.2 49.55 45.54 3.56 198.8 5.07 2,629
Sawarda Oct'21 444.2 434.4 415.7 18.7 168.87 160.70 1.53 389.0 2.79 2,228
Sawarda Nov'21 396.2 404.7 404.7 0.0 139.85 106.70 1.48 266.5 3.77 2,093
Sawarda Dec'21 330.5 341.2 341.2 0.0 115.74 105.36 1.30 89.3 3.18 2,025
Sawarda Jan'22 360.1 380.9 348.2 32.7 137.46 136.12 1.06 239.9 2.67 2,268
Sawarda Feb'22 367.7 351.2 351.2 0.0 126.46 125.65 0.79 245.5 2.94 2,201
1 DG idling is taken when Watt Hours supplied is zero while DG ignition is ON. Excludes missing watt hour periods

0.5.1 Daily kWh trend

The underlying data used for all the above reports can be seen in the following daily energy plots. The light blue solid line is the statistically interpolated EB kWh data, red line is the DG kWh and the dark blue line at the top is the total AC kWh (sum of EB and DG kWh). The emphasized dark blue smoothed line with a grey envelope shows the 95% confidence interval of the total AC kWh (total site_name energy consumption) trend.

Note: the light grey dashed lines inside EB and DG line charts are from the raw data if no interpolation was performed.

Note: Underlying raw data in csv format can be supplied for all the above charts.

0.6 DC Load

DC load contributes the biggest energy consumption for any BTS site. The load patter for the full day is different for each site and may depend upon the radio equipment that reduces load based on the traffic.

0.7 Missing Data

0.7.1 Energy Meter missing data

Missing data is an important metric for the overall quality of the IOT project. We should strive to get this value to zero for all sites. The following plot shows the trend in hours of data missing per month for each of the sites that are color coded. The trend shows a strong and drastic reduction in outages coinciding with renewed interest in the project from all concerned in January.

0.7.2 Fuel level missing data

When we examined missing data for fuel, the top fuel sensor outage occured for the site : Jobner in the month Dec’21 for a total period of 312 hours. There was a net fuel level difference across this outage period starting on Dec 18 of 158.1 litres.