As the rain usually isn’t falling when it is needed, some form of storage is required. The size of the storage combined with the water demand and rainfall are responsible for the reliability of the system. The usual form of storage is a tank, which must be relatively watertight (say, leakage of less than 5% of daily abstraction) and hold the required volume. Other requirements include:

• The ability to be able deal with excess input by overflowing in manner which doesn’t damage the tank or its foundations
• Exclusion of vermin and mosquitoes
• Exclusion of light (so that algae do not grow and larval growth is inhibited)
• Ventilation to prevent anaerobic decomposition of any washed in matter
• Easy access for cleaning
• Sufficient structural strength to withstand wear and tear, and occasional large natural forces
• No hazards to passers-by or small children
• Not giving the water an unacceptable taste

Geometry

Material economies can be made on water tanks by considering the geometry of surface area to volume. Table 4 shows the relationship between various shapes and their respective material economies.

Table 4: Idealised tank shapes

Shape	Material Penalty	Notes
Sphere	1.0	Perfect spheres are only possible underground or partly underground however the shape can be approached using doubly curved structures Good stress characteristics with little bending stress All doubly curved structures need great skill or excellent tooling (or both) to manufacture reliably Only suitable for mouldable materials such as cement and clay or flexible materials such as some textiles and plastic sheeting
Cylinder	1.2	The most popular shape for water tanks Hoop Stresses are efficiently accepted, however a fixed joint between the tank wall and base will cause bending and sheer stresses near the joint Suitable for use with ether mouldable materials or materials which can be bent on one direction (such as metal sheet)
Half Sphere	1.3	A popular shape for underground tanks as the pit is easy to excavate and it is believed to have good material economies Requires a large, free standing cover Underground tanks are simple to make with this shape using mouldable materials
Cube	1.4	Perfect spheres are only possible underground or partly underground Bending stresses are high toward the corners Very simple to construct using familiar house building techniques Suitable for all materials including bricks and blocks

The aspect ratio can also have an effect on the cost of the tank. As the tank differs from the idealised shape material economies suffer. Figure 5 shows this relationship for tanks of constant section such as cylinders and cuboids. As can be seen the material diseconomies are stronger in short, wide tanks than in long thin tanks. This is a distinct advantage, as many users prefer to have tanks with a smaller footprint, as land is often at a premium.

Figure 5: Effect of Aspect Ratio on the Material Economies of Tanks of Constant Section

Materials and techniques

Precast Concrete

In some developed countries such as Australia and Germany precast concrete tanks form a large part of the market. The tanks are cast in sizes up to 35m3 under controlled conditions, delivered to the site by truck and simply installed. The economies inherent in this strategy revolve around the ability for the factory to specialise in this type of construction, the use of appropriate jigs and the ease of installation reducing on-site labour costs. In Germany most tanks are sited underground to reduce space requirements.

There have been several attempts to build such tanks in developing countries such as Brazil (Szilassy, 1999) and Kenya (Lee & Visscher, 1990) using shuttering with corrugated iron, however the technology has generally proven too expensive to be widely replicated. Precast rings have been used successfully in Bangladesh (Ferdausi & Bolkland, 2000) already being produced in quantity for well lining. This ability to mass produce items gives the technique some promise in the field of tank components such as segmented covers and filter boxes and concrete is often used for ancillary work around tanks such as foundations, drainage and soakaways.

Steel

Steel tanks of various sizes have been used throughout the world for many years and are still popular today. They range from the ubiquitous steel drum found outside almost every house in East Africa to gigantic 1.5Ml structures used to supply remote communities in Australia. The tanks can be delivered to a site and installed in a short time by a skilled person. An extremely firm foundation is often not required, as the steel structure will "give" a little to accommodate any settling.

In developing countries problems with corrosion on the bottom of the tank have been observed after about two years service. Building a concrete ring around the base of the tank can repair this, however such failure in the field has limited the steel tank’s acceptance and wider application. The problem does not generally appear in tanks built in developed countries as steel tanks are generally either coated with plastic on the inside of the tank (BHP Pty. Ltd., 2000) or lined with a plastic composite bag (Pioneer Tanks Pty. Ltd, ).

Oil drums are one of the most widely dispersed water containment stores in the world. However, they have unique problems due to their previous use.

• Most drums now used for containing water have previously contained chemicals, many of which are toxic.
• They have also usually been opened in such a way that they are uncovered and thus present an ideal environment for mosquito breeding and yield low quality water.
• Water extraction can be a problem.

If these problems can be solved inexpensively, then drums form a readily available supply of inexpensive (if small) storage units.

Plastic

Plastic tanks, usually made from HDPE or GRP form the fastest growing segment of storage provision. They are already popular in developed countries where they compete directly with older technologies such as steel or concrete on a direct price basis. In developing countries, these tanks are generally more expensive by a factor of 3-5 which has slowed their adoption, however this is changing, In Sri Lanka the price penalty of a plastic tank is down to about 1.5-2 and in South Africa they are generally considered cheaper (Houston, 2001)

Even in countries where there is a price premium for plastic tanks, they are often employed by water supply organisations, as they are quick to install and are known to work reliably (usually backed by a manufacturers guarantee). Consumers also like the tanks and see them as the most up-to-date method of storing water, however problems of cleaning and the water heating up in the black tanks have been identified.

An application of plastics that is highly cost effective is the use of plastic lining materials with otherwise local techniques. An example of this is the tarpaulin tank originally used by Rwandan refugees in southern Uganda and subsequently improved by ACORD and widely replicated (Rees, 2000). The cost of the tank is roughly ¼ of an equivalent ferrocement tank. The frames of the tanks are, however liable to termite attack and the tarpaulins themselves have failed in service in some areas reportedly also due to termites although this has not been proved out and contrary stories exist.

Ferrocement

Ferrocement is the technology of choice for many rainwater harvesting programmes, the tanks are relatively inexpensive and with a little maintenance will last indefinitely. The material lends itself to being formed into almost any shape and outside tank construction is used for boat building and even sculpture. It has several advantages over conventionally reinforced concrete principally because the reinforcement is well distributed throughout the material and has a high surface area to volume ratio.

• Cracks are arrested quickly and are usually very thin resulting in a reliably watertight structure
• It has a high tensile strength (in the region of 0.8-1.5 Mpa before cracking)
• Within reasonable limits, the material behaves like a homogeneous, elastic material

The technique was developed in France in mid 19th century and was initially used for pots and tubs and even boats, (Morgan, 1994) but was transplanted by less labour intensive reinforcement methods. Tank construction with ferrocement has been ongoing since the early 1970s (Watt, 1978) was popularised by its use in Thailand (IDRC, 1986) and has spread to Africa (Nissen-Petersen & Lee, 1990), South America (Gnadlinger, 1999) and Sri Lanka (Hapugoda, 1995) among others. Tank construction using ferrocement involves the plastering of a thin layer of cement mortar (typically 1 part cement to 3 parts sand mixed with about 0.4 parts water) onto a steel mesh.

The method of construction involves the plastering of a thin layer of cement mortar (typically 1 part cement to 3 parts sand mixed with about 0.4 parts water) onto a steel mesh. Despite being described as a "low skill" technique, workmanship is strong issue with all ferrocement constructions. The thickness of mortar is often as little as 5mm giving little room for error when covering the mesh. A good former can reduce error and is often the reason behind successful designs. Increasingly though, these formers are being abandoned due to their cost and to gain flexibility in size. The mesh itself is then used as a base to cement on to, but these tanks usually have a consummate increase in wall thickness and thus cost.

The most popular design of ferrocement tank continues to be the straight cylinder. Formers are easy to construct using sheet metal or BRC mesh, and there are usually no foundation problems as the base is wide enough. There can be some problems of cracking at the base if stress concentrations are not accounted for in the design and there have been some reports of cracking at the lid interface. To combat this several designs such as the Sri Lanka "pumpkin" tank (Hapugoda, 1995) have been produced with a rounded shape to break up the sudden junctions.

Even more popular than cylinders, but not technically "ferrocement" is so call "Thai jar", which has been mass-produced in Thailand for about 20 years. There are more than 14 million of these jars throughout Thailand with capacities ranging from 0.5m3 to 3m3 (Bradford & Gould, 1992)The jars are manufactured by plastering mortar with no reinforcing onto a mould in the shape of the jar. The moulds are centrally produced to a low cost and the jars are made in reasonable numbers in small workshops(Ariyabadu, 2001).

Another method of employing mass production techniques is to make the tank in sections. This has been applied in India at the Structural Engineering Research Centre (Sharma, 2001) where the tank is made in either full height or half height segments on-site or in a central location. The segments are then shipped out by truck joined together on-site in a single day. The segments themselves have a much-reduced thickness as they can be made horizontally at a comfortable height on well-designed jigs. Segmented techniques have also been tried in Brazil (Gnadlinger, 1999) with the segments made on-site. The material cost is similar to same sized ferrocement tanks made on a former but the tanks are quicker to build.

The realm of ferrocement has also seen several attempts to reduce cost by replacing metal reinforcing with other materials such as bamboo and hessian. Although there have been some success stories (Sharma & Sen, 2001) there are a number of notable and large-scale failures. In Thailand, over 50,000 bamboo-cement tanks had been built when a study by Vadhanavikkit and Pannachet. (Vadhanavikkit & Pannachet, 1987) revealed that fungi and bacteria were decomposing the bamboo and within a year the strength of the reinforcing had reduced to less than 10% and some had rotted away altogether. The study concluded that the majority of bamboo cement tanks would fail, some suddenly and dangerously. Another programme in East Africa by UNICEF and Action Aid in the 1970’s developed the "ghala basket" an adaptation of a traditional grain basket made waterproof by the addition of mortar. By the mid 1980s, it was becoming clear that these baskets were susceptible to rotting and termite attack and the design was abandoned (Gould, 1993).

Bricks

Bricks and blocks of various types are widely used for building in many developing countries. The materials are found locally and local people prepare the bricks themselves, thus the cost of the bricks is usually low and all monies remain in the local economy. They can be made from a number of different materials such as burned clay, cut stone, soil stabilised with a small amount of cement or even concrete. Unfortunately, while bricks are useful for building work they less well suited to tank construction. As they have a poor strength in tension and so the tensile forces are usually taken up by the mortar and by adhesion between the mortar and bricks, which is usually fairly low. Brick tanks can also suffer a cost disadvantage as the thickness of the tank is set by the width of the bricks and if the bricks are poorly fitting, such as when making a cylinder of small diameter, they can demand more cement than an equivalent ferrocement tank.

Interlocking bricks, usually using stabilised soil, have been tried in several places including, Thailand (IDRC, 1986) and Uganda (Rees & Thomas, 2000) (New Vision (Kampala), 2001). A machine for making interlocking Mortar blocks is also available (Parry, 2001). Most of these designs interlock vertically relying on sheer between the mortar and block to take the stresses. A more satisfactory solution would be to interlock the block horizontally on top and bottom surfaces, however this does not appear to have been investigated.

Underground or aboveground

The simple gambit of building the tank under the ground is a popular method of cost reduction in tank building. Foundations problems are avoided completely as the tank is immersed in the supporting soil and so very large tanks can be constructed with relative ease. Nissen-Petersen (Nissen-Petersen & Lee, 1990) has developed a 90m3 tank in Kenya which has proved popular for schools and public buildings.

Of more interest to the field of VLC, tank building is the property of stable soils to reliably take the force of the water meaning that any cement or render is needed only as a sealant. Thomas and McGeever (Thomas & McGeever, 1997) have made several tanks in West Uganda using a 25mm layer of mortar applied directly to the soil with few reported problems after 5 years service. In Ethiopia, a number of tanks have been made using a similar technique with a soil-cement (Hune, 2001), further reducing the requirement for imported material.

In the northern China, the soil is so stable that people simply dig their houses out of the earth. Here, a bottle shaped store known as a "shuijao" has been used for centuries. The ground was simply dug out and mud compacted onto the walls. Recently, the government has been improving these tanks using cement lining for improved water retention (Zhu & Wu, 1995).

Failure of underground tanks can be a problem, leaks are difficult to locate and equally difficult to repair. In a study by Ranasinghe (Ranasinghe, 2001), below ground brick tanks were found to have been holed by tree roots resulting in losses of up to 2.5 litres per day. The other major failure of underground tanks is by the water table raising and empty tanks "floating" out of the ground (Joy, 2001) or simply collapsing under the strain of the outside water (De silva et al., 2001).

Tanks lining the ground with plastics have been tried since the 1970s (Maddocks, 1975), often with little success, however designs such as the Ugandan Tarpaulin tank (Rees, ) and the common use of polythene lining for reservoirs and ponds, (Santvoort, 1994) show that the method can be used with careful design. The usual failure modes are tree roots as with other underground tanks, ultraviolet degradation and vermin intrusion. The tanks are, however immune to floatation as they simply flex out of the way. Reports of termites attacking underground plastic sheet tanks are common, however there are equally reports of termites living under the plastic and not damaging it. The matter of UV degradation should be neatly avoided in an underground tank, as given that an appropriate lightproof lid is a requirement for health reasons, the plastic itself should not be exposed to sunlight.

A very inexpensive method of storage is to simply use the groundwater table. In cities such as Delhi, citizens are being encouraged to divert the water from their roofs into the ground to "recharge" the groundwater (Ranade, 2001). As well as generally improving the groundwater level, there is a localised effect whereby the water forms a "mound" under the recharge point creating a nominal private store (although the water will eventually travel outward).

A summary of the advantage and disadvantages of underground and overground storage are shown in Table 5.

Table 5: Pros and Cons of above ground and underground storage

	Pros	Cons
Above ground	Allows for easy inspection for cracks or leakage Water extraction can be by gravity and extraction by tap Can be raised above ground level to increase water pressure	Require space Generally more expensive More easily damaged Prone to attack from weather Failure can be dangerous
Underground	Surrounding ground gives support allowing lower wall thickness and thus lower costs More difficult to empty by leaving tap on Require little or no space above ground Unobtrusive Water is cooler Some users prefer it because "it’s like a well"	Water extraction is more problematic - often requiring a pump, a long pipe to a downhill location or steps Leaks or failures are difficult to detect Possible contamination of the tank from groundwater or floodwaters The structure can be damaged by tree roots or rising groundwater If tank is left uncovered children (and careless adults) can fall in possibly drowning If tank is left uncovered animals can fall in contaminating the water Heavy vehicles driving over a cistern can also cause damage Cannot be easily drained for cleaning

Overall, about 80% of users express a preference for overground tanks despite a cost penalty of 50%.