Pitfalls on the road to urban grid independence

August 11th, 2014, Published in Articles: Energize

 

Urban grid defection, also known as grid-independence or off-grid generation, is a topic which has received much attention of late. There are numerous articles suggesting that, in the not very distant future, it will be economically possible for urban consumers to operate independently of the grid with a suitable combination of renewable energy sources and storage. Most claims seem to have neglected the potholes and chasms in the road to self-sufficiency, either deliberately or in error by making use of poor information. This article highlights some of the problems which potential grid defectors will be faced with.

Rooftop solar is becoming more and more popular worldwide. Current claims are that in the foreseeable future the residential distribution grid will be obsolete and that all residences could be powered from renewable sources [1]. The claims propose that with a well-proportioned combination of solar PV and storage, residences will not require any electricity from the grid, but may, as is the case in countries like Germany, sell power into the grid.

This applies to residential, as well as commercial and industrial properties. What we are looking at here is urban off-grid, which has more stringent requirements than urban grid-connected and is vastly different from remote off-grid, or mini-grid operation. The driving force behind this is the belief that the cost of powering an urban installation using hybrid PV plus a storage system will become cheaper than the cost of utility or grid power, while still providing the same power availability as the grid.

Current thinking on urban off-grid systems postulates a hybrid solar PV/ battery system which will allow energy to be stored in the battery during the day and used to power the installation at night. Battery storage is seen as the bridge over the gap that exists when the solar PV does not generate power at night.

System sizing

The PV system will have to generate sufficient energy to supply the consumer’s needs on a daily basis. Assuming that the maximum power consumption is P kWh and an equal consumption between daylight and night time, with an efficiency of 80% the solar PV system will have to provide P/2 + P/(2*0,8) = 1,125*P kWh. And the size of the storage battery will be P/2 kWh.

The simple picture does not take into account the fact that power levels from the solar system are not constant but vary with the time of day, so the storage system will have to be dimensioned to cater for early morning demands on the system when the sun is weak.

A system of this sizing will work fine provided the sun shines every day, which it doesn’t. What will happen to the urban off-grid consumer on a day when the solar radiation falls to half it’s normal value? Only 0,5 P kWh will be generated and the power available will also fall to half that required. Clearly more storage is required to cater for lower than normal radiation levels.

How much low radiation can be expected ?

Most databases give long term average figures. Some may even give statically distribution curves such as the ever popular Weibull function which show the probability of occurrence of different levels of solar radiation. Some solar designs may take the occurrence of low radiation days into account, but most, in an attempt to cut costs, will avoid such issues. This is not a problem if the system is connected to the grid, but if the consumer has defected from the grid, then this is a problem.

Fig. 1: Consecutive days of low radiation can be a  problem for an off-grid system.

Fig. 1: Consecutive days of low radiation can be a problem for an off-grid system.

Many of the studies done in connection with this topic make use of monthly or even yearly aggregated figures for solar radiation to calculate costs. Statistics and distribution curves give the false impression that low radiation days occur singly and are evenly distributed. Unfortunately nature is not as benevolent as this, and delivers days like that in clusters rather than individually, forming gaping chasms which will swallow any off-grid consumer if not taken into account. The chasms that need to be avoided are illustrated in the example shown in Fig. 1, which is the normalised annual radiation pattern for a South African site within an area regarded as a high annual radiation area [2].

Besides the minor variations in radiation, the pattern in Fig. 1 shows several groups of consecutive days of poor radiation, which will have a major impact on the capability of the system to provide reliable power. Averaged annual radiation figures and distribution curves give no indication of these events.

Case study

Fig. 2 shows an example of daily solar radiation from a South African Weather Service (SAWS) station located in an area regarded as a good insolation area. The site receives steady irradiation for 10 days before hitting a four-day chasm.

Fig. 2: Days of low radiation occur in clusters as well as singly.

Fig. 2: Days of low radiation occur in clusters as well as singly.

Table 1 summarises the energy balance for this period.

Table 1: Energy balance for the first event. 

Day

Radiation level %

Deficit %

Running total deficit %

9

100

0

0

10

95

5

5

11

48

52

57

12

55

45

102

13

68

32

134

14

80

20

154

15

100

0

154

16

100

0

154

Over the four day period the system will be short of 154% of its energy requirements, if the system is sized based on daily requirements. To survive, additional storage, in this case of 1,5 days energy, will be required, in addition to the storage required for daily operation.

Low radiation days not only occur in clusters but can also occur in consecutive clusters separated by short periods of normal radiation. To see the effect of this, consider what happens if the necessary 1,5 days of stored energy has been provided for the first event and this is followed shortly by a second event, as shown in Fig. 3.

Table 2 summarises the energy balance for the second event.

Table 2: Energy balance for the second event. 

Day

Radiation level %

Deficit %

Running total deficit %

15

100

0

0

16

100

0

0

17

80

20

20

18

93

7

27

19

80

20

47

20

62

38

85

21

60

40

125

After two days of full sunlight the system hits another low radiation cluster with an energy deficit of 125%. The system has 1,5 days of backup storage, but this has been depleted from the previous event. Energy to recharge the storage batteries can only come from the PV system which will have to be oversized to allow for recharging the back-up storage. In this case there are only two days available to recharge the storage, so the PV array will have to be oversized to provide the extra energy over two days to cater for the second event. Oversizing decreases the deficit on the low radiation days, in addition to providing for recharging the storage batteries. Table 3 shows the effect of oversizing at 20% on the performance of the system during these two events.

Table 3: Effect of oversizing a PV array by 20%. 

Day

Radiation level %

System output

Surplus

Storage level

Cumulative deficit

1

100,0

120,0

20,0

20

0,00

2

99,4

119,3

19,3

39,3

0,00

3

96,4

115,7

15,7

55,0

0,00

4

99,2

119,0

19,0

74,0

0,00

5

98,1

117,7

17,7

91,7

0,00

6

99,7

119,6

19,6

111,4

0,00

7

95,1

114,1

14,1

125,5

0,00

8

93,0

111,6

11,6

137,1

0,00

9

100

119,9

19,9

150,0

0,00

10

95

113,9

13,9

150,0

0,00

11

48

57,0

-43,0

107,0

43,05

12

55

66,8

-33,2

73,8

76,20

13

68

81,6

-18,4

55,4

94,56

14

80

96,2

-3,8

51,7

98,31

15

100

120,0

20,0

71,7

0,00

16

100

120,0

20,0

91,7

0,00

17

80

96,2

-3,8

87,9

3,78

18

93

113,8

13,8

101,7

3,78

19

80

96,0

-4,0

97,6

7,83

20

62

76,2

-23,8

73,9

31,59

21

60

71,6

-28,4

45,5

59,95

Increasing the size of the array by 20% reduces the deficits from 160 to 98% and 125 to 59%, and the additional 20% allows recharge of the storage batteries within the days available, and the storage of 1,5 days is adequate.

Actual sizing will require an analysis of multi-day events from historical data. Final design will be a trade-off between the cost of the array and the cost of storage. One could consider putting in more storage than necessary to cover the short separation between events without recharge, and only increase the array by the small amount required to recharge the storage, but again this will be decided by the recharge time available. Restrictions in space may dictate the maximum size of the array.

This is an extreme case and not all low radiation events are so closely spaced. Fig.1 shows periods varying between 4 and 14 days between events. Irrespective of this, the sizing of the array will have to include a recharge capability, and the more severe the low radiation pattern is, the larger the recharge allowance will have to be. The actual size can only be determined by a day by day analysis of data for the area, and cannot be resolved by mathematical or statistical models based on averaged data. The recharge ratio, i.e. the ratio of the recharge capacity to the basic system capacity, is set at between 30 and 50% in high reliability telecommunications power supplies. The data set for the chosen site exhibits 18 such multi-day events, in fact they are much more prominent than single day events, which tend to be small deviations from the maximum.

Fig. 3: Solar radiation pattern for three weeks.

Fig. 3: Solar radiation pattern for three weeks.

Potential problems with commonly used designs

Problems arise due to the reliance of studies on long term average data and poorly understood statistics. Long term averaging tends to mask the occurrence of multiple occurrences, and statistical distribution charts, no matter whether Weibul or other distribution functions, cloud over the fact that such things do exist. No distribution chart that I know of will indicate the occurrence of four days of low sun followed by two days of sun and another three days of low sun. Distribution functions create the false impression that low radiation days are conveniently evenly spread over the time period. Most will simply give an indication of the possibility of occurrence of poor days, and none give an indication of the pattern in which they will occur. The mistake is often made of assuming that the low probability of occurrence of an event means that it will not occur. Data used to derive probabilities comes from events which did occur, and assuming that past weather records are a good indication of future behaviour, will occur again.

South African prospects

How does the rest of South Africa fare in terms of bad sun days, and what potholes can the prospective off-grid home-owner expect to encounter? A study of cloudy and sunny days has been undertaken by SAWS [3], which provides an insight into sunshine and cloudiness in South Africa. The study is based on records of sunlight hours at stations around the country and not on solar radiation. Measurements made using a Campbell-Stokes sunshine recorder, which records sunlight hours but not actual radiation levels. Bear this in mind when looking at the results.

Fig. 4: Average annual occurrence of different classes of sunshine duration days [3].

Fig. 4: Average annual occurrence of different classes of sunshine duration days (SA Weather Service [3]).

Fig. 4 shows the average annual occurrence over the country of different classes of sunshine duration days, which are classified as follows:

  • Overcast days: No sunshine or total cloud cover for the total duration of the daylight time. This varies from 30 days/annum in the east to less than 1day pa in the far west
  • Cloudy days: Up to 10% of the possible sunshine hours received every day. This varies from 50 cloudy days in the east to less than 10 in the west
  • Sunny days: At least 50% of the possible sunshine hours. This varies from 240 in the east to over 330 in the west
  • Fine days: At least 90% of possible sunshine duration

Table 4 shows the average monthly frequencies of cloudy days (less than 10% of sunlight hours – this again is a long term average and covers from 0% to 10% sunlight hours) [3].

Table 4: Average monthly frequencies of cloudy days. 

Location

January

February

March

April

May

June

July

August

September

October

November

December

Bloemfontein

1,2

1,7

1,8

1,1

0,6

0,6

0,5

0,5

0,8

0,9

1,0

0,9

Cape town

0,5

0,5

0,6

1,6

2,3

3,6

3,2

2,8

2,0

1,2

0,8

0,8

Durban

7,3

5,1

4,4

3,6

2,2

1,8

1,7

3,1

6,8

8,2

8,1

6,6

East London

5,2

4,8

5,4

3

1,8

1,4

1,2

2

3,8

5,6

5,2

4,8

Estcourt

4,1

3,6

3,3

2,5

1,2

1,1

1

2,1

3,6

5,9

5,1

3,9

George

4,8

4,2

4,6

3,0

2,5

2,5

2,6

2,7

3,8

5,0

4,4

3,5

Johannesburg

2,2

2,2

2,3

1,5

0,8

0,6

0,5

0,7

1,2

1,8

2,1

1,8

Kimberly

0,8

1,0

0,9

0,7

0,3

0,4

0,5

0,6

0,4

0,7

0,8

0,9

Mafikeng

1,2

1,8

1,9

0,9

0,8

0,4

0,6

0,3

0,8

1,2

1,8

1,2

Polokwane

1,6

1,7

1,9

1,2

0,6

0,4

0,6

0,8

1,2

1,6

3,0

1,6

Port Elizabeth

2,8

2,9

3,8

2,0

2,0

1,9

1,4

2,1

3,0

3,9

2,8

2,4

Pretoria

2,1

2,1

2,9

1,1

0,2

0,3

0,2

0,4

1,1

1,3

1,8

1,6

Upington

0,6

0,4

0,7

0,7

0,3

0,2

0,2

0,2

0,3

0,4

0,3

0,6

The figures given above only represent sunlight hours and not actual radiation. The effect on radiation figures will depend on when the cloudy period occurred. Records show that there is a variation in diurnal cloud cover patterns in different parts of the country [3]. In Durban for instance we can expect, on average again, eight overcast days per month, or 25% of the time, in summer. There is no indication of the pattern of occurrence but experience suggests that this would include several days of consecutive cloud cover.
These figures are generally not reflected in solar radiation statistics, and represent a challenge to the urban off- grid householder. Remote off-grid sites and mini grid systems overcome this problem by incorporating fossil fuel powered generators, such as diesel generator sets, to provide back-up for long periods of low radiation. This would probably prove impractical for the urban consumer, and certainly increase both the capital and operating costs of an urban grid independent system. For most  urban consumers, going off-grid will require more than a full day’s energy storage, over and above the operation requirements of a PV system, and an oversized array to allow for recharge of this storage. On the east coast longer storage may be required, as more overcast days imply longer groups of consecutively bad sun-days, although there are no statistics on this.

What does all this mean?

Simply that an urban off-grid system which has the same reliability as the grid over the long term is going to cost a lot more than a grid connected system, and the claims of grid parity are in fact only applicable to grid connected systems. Concern from utilities over this may well see a premium attached to grid connection, to cover the cost of maintaining a generation, transmission and distribution network to provide power when the sun doesn’t shine.

As the driving force behind this movement to go off-grid seems to be the concept of grid parity, or that electricity can be produced on site cheaper than from the grid. Consideration of the above factors shows that urban off-grid can be a lot more expensive than urban grid connected. Urban off-grid cannot be compared with remote off-grid, which often consists of hybrid systems, the cost of which can be justified either by the fact that no grid connection is available, or the cost of bringing the grid to the site exceeds that of the system. The argument in favour of urban grid defection seems to be ill founded and lacking in consideration of all the facts, and may be a case of enthusiasm overtaking good sense.

Worst case or average ?

Another issue which is often not taken into account is the seasonal variation of solar radiation. For an off-grid system, which will probably have a fixed elevation, the difference between summer and winter PV output could be as much as 50% for non-optimal tilt angles. Many of the studies supporting grid defection use annual average figures and neglect the fact that the solar array will have to provide the full energy requirements during the worst month, which means a larger array and higher cost. Most grid connected systems will use annual average radiation figures for sizing.

Fig. 5: Variation of solar radiation on horizontal surface over full year [2].

Fig. 5: Variation of solar radiation on horizontal surface over full year [2].

Fig. 6: Variation of solar radiation on tilted surface over full year (using optimal tilt angle for every month).

Fig. 6: Variation of solar radiation on tilted surface over full year (using optimal tilt angle for every month).

Figs. 5 and 6 show the variation in radiation for a horizontal and tilted surface for the station used in this article. The optimal angle [4], for each month was used to adjust the radiation values.

Conclusion

As attractive as grid defection may initially appear, it is probably not a good option for urban domestic, commercial and industrial consumers. Solar radiation patterns are not regular and smooth but exhibit groups of consecutive days of low radiation. The pattern of such groupings is not evident from statistical data and needs to be determined by examining data records. This will result in an increase in the capital cost of the installation as more storage, or more PV panels, or both may be required than originally thought to provide power through periods of low radiation. The actual recharge ratio will be determined by a trade off between storage size and PV costs. Irrespective of what final configuration is chosen, the ultimate cost of grid independence may be considerably more than the cost of grid electricity.

References

[1]    P Bronski, et al: “The economics of grid defection”, Rocky Mountain Institute, February 2014.
[2]    South African Weather Service: “Daily radiation data for De Aar BSRN station”, 2005.
[3]    A Kruger, D Esterhuyse: “Climate of South Africa: Sunshine and cloudiness, WS46”, South African Weather Service, 2005.
[4]    A Eberhard: “A solar radiation data handbook for Southern Africa”. Energy research institute, University of Cape Town, 1990.

Send your comments to: energize@ee.co.za

Related Articles

  • Analysis: The next steps of the Eskom restructuring and turnaround plan revealed…
  • Managing and operating solar assets: Five key considerations
  • Support a budding scientist and help build a skilled South Africa
  • Invitation to attend SANEA’s Carbon Tax Colloquium
  • SA biogas conference generates renewed interest