Skip to main content

DFID KaR R7833: Roofwater Harvesting for Poorer Household in the Tropics

1 December 2000 - 31 March 2003

The Roofwater Harvesting for Poorer Household in the Tropics project is aims to extend the knowlege of roofwater harvesting and to develop and test roofwater supply systems affordable by the poor. The project is a partnership between the DTU and three other national organisations:

Development Technology Unit

School of Engineering,
University of Warwick,
Coventry CV4 7AL, UK

Tel: +44 24 7652 2339
Fax: +44 24 7641 8922

Contact: Mr. Brett Martinson



ACORD Mbarara
PO Box 1394

Tel: +256 41 267667/8
Fax +256 485 213041

Contact: Mr. Dunstan Ddamulira


Lanka Rain Water Harvesting Forum 

239, High Level Road
Colombo 5
Sri Lanka

Tel: +94 75 524612/3
Fax: +94 1 853053

Contact: Ms. Tanuja Ariyananda


Water Action

PO Box 13367
Addis Ababa

Tel: +251 1 61 42 75
Fax: +251 1 66 16 79

Contact: Mr. Girma Mengistu



Goal, Rationale and Outputs


To raise the well-being of the rural and urban poor in the tropics through improved water supply.


The development and assessment of very low cost domestic roofwater harvesting (VLC DRWH) technologies to meet the water needs of poor households in the tropics.

Programme Rationale

Domestic roofwater harvesting is practised in many countries, but rarely on a significant scale and predominantly in one of two forms. 'Informal' DRWH is cheap but produces small amounts of not-very-clean water. Formal DRWH is promoted (with subsidies) by rural water agencies or adopted privately by middle class families - it employs fairly large water stores and attempts to satisfy the bulk of a household's water needs. Between these two forms can be placed 'very-low-cost' partial RWH, that is designed to meet a significant part (say 65%) but not all of domestic water need at a system cost affordable by even relatively poor households. Such VLC DRWH, the focus of this research, is only achievable under certain (humid) climatic conditions, however the global population living under such conditions exceeds 800 million. It is however rarely met with because the relevant equipment has not been well developed, such equipment as is used performs poorly, partial domestic water sources although widely used are not considered normal by water agencies and finally because DRWH represents a form of self-help that does not fit with either communal or commercial models of water supply.

In many countries however, RWH is becoming better known and better thought of. In a very few countries it is already apparently practised by over 10% of households. In order to assess the potential for a considerable expansion of DRWH usage by the group generally most in need of water, the poor, we have to ascertain the underlying economic viability of this mode and understand impediments to its use. We have to demonstrate how far it can give potable, reliable and of course affordable water. To this end improvement of the technology is required, driven by a clear understanding of the specification it needs to meet. Unfortunately little is reported about such cheap small systems and even the literature on larger systems is inadequate to reassure water policy makers or potential system builders.

The programme therefore has two interleaved parts. One is to improve understanding of the social impact, potential and performance of partial RWH as practised by families in small houses, assessing its cost and benefits. The other is to improve the DRWH technology itself, adapting it to this VLC role and market.

The potential stakeholders in this technology are the direct beneficiaries (households, the artisan installers of systems, manufacturers of components) and a range of intermediary organisations (NGOs and CBOs engaged in poverty-reduction programmes, public health and water agencies, development funding agencies etc.). As a research programme the outputs, although generated through interaction with representative members of the first group, have mainly to be directed towards the second group.


Research Methodology

The project divides into three main threads: technical, socio-economic and policy. Each thread will be managed independently with regular formal cross-reporting (as well as contacts through less formal channels). The project divides into three main stages. Stage A is the Initial Analysis, Stage B is Design and Stage C is Assessment.


The technology thread forms a product development cycle. The product can to some extent be broken down into autonomous components such as gutters, inlet filters, tanks and water-management aids. It is a generic product range, rather than one to be manufactured by a single enterprise and in consequence quality assurance has to be 'designed in' rather than wholly left to the individual producer.

In Stage A (Initial Analysis) technology research will take the form of fieldwork to ascertain the technical constraints and range of materials and techniques available. PRA techniques will be used along with market surveys of available products and materials. This will be combined with laboratory work to generate general solutions to design problems and appropriate material specifications. The initial technical analysis will be combined with initial socio-economic analysis to form the specification for the Development/Innovation stage.

Stage B, Development/Innovation and Technical Assessment will commence in UK then transfer to Kandy, Sri Lanka for iterative building and testing taking 4 months of rainy season. Promising solutions will also be tried in other partner countries and adjusted to match their local conditions, knowledge and expertise. Prior work indicates that the areas in most need of innovation are assuring high water quality in small systems, fitting systems to tight architectural constraints in poor urban housing, increasing hardware durability and minimising losses prior to water entering storage.

In Stage C the final designs will then be user-tested under a representative range of conditions at urban and rural sites in the three partner countries. Because the systems are quite cheap to install (and some cost-recovery will be employed) at least 150 will be built, a number large enough to permit considerable range of sites to be tested without sacrificing statistical robustness. However the time-scale of the programme precludes realistic trials of durability, except where (e.g. with cistern handpumps) accelerated testing can be employed.


The key social issues and gaps in the economic data will be addressed in the Main Phase through two socio-economic and gender studies. The first will be conducted during Stage A ('Initial Analysis') in the first year of the project. It will identify principal socio-economic and gender issues influencing the adoption and use of DRWH systems. It will also collect such economic data such as willingness to pay, extent of 'informal' RWH, fraction of hard roofs and seasonal variations in water costs. The studies will use a combination of secondary sources, revealed preference or contingent valuation surveys, and rapid appraisal methods in communities and individual households.

The second study falls in Stage C of the project and will assess the social and gender impacts of technical innovations following their implementation in study communities. It will focus on the implications of the new designs and of the adoption of permanent RWH at household and community levels: examining for example, who benefits, who loses and whether others follow suit spontaneously. The economic measurements will be extended and some measurements will be repeated to identify strong short-term trends. It is probably that users' reactions will change over time since first encountering VLC DRWH. We shall therefore employ measurements at both (the few) sites where such systems are of long standing and at sites where we have deliberately introduced changes. Indeed, we shall have to some degree to separate study of the social impact of introducing VLC DRWH from the study of economic responses to design refinements. The former will take advantage of the 15 plus months available after the building of more conventional systems during Stage A. The latter will have to rely on fairly immediate responses to the more novel systems built during Stage C.

The two socio-economic studies will be designed in collaboration with the partners. The design of the main social and economic survey to be conducted during the first year of the project will be consolidated following the Inaugural Meeting to be held in Sri Lanka in July 2001. The process will take account of the findings from the field survey conducted during the Inception Phase, coupled with other experiences of partners, and the literature review.

This participatory approach will ensure there is a common understanding between partners with regards to the purpose of the survey and the methods of data collection.


Work on the policy thread will be ongoing throughout the project. The primary vehicles for data gathering on policy will be a documentary review of current and historical material to provide comparative international data on the direction and speed of change in attitudes towards DRWH. This will be backed by direct interviews with water agencies in partner countries and with members of water development NGOs at conferences and other international functions.

Site Selection

In the selection of both sites and survey details, there is a conflict between being comprehensive and being representative. Sites chosen to maximise the range of scenarios are unlikely to give statistical depth to the commonest scenarios. Moreover access to sites is constrained by the field contacts of our partners and the obtaining of official permission to research in sensitive locations - for example some urban settlements. This issue is to be discussed at the key Inaugural Meeting. Within the constraint of having not less than 5 similar systems at any one site, the experimental design will maximise site types. It is thought to be unrealistic to formally extrapolate findings to regional estimates of such parameters as fraction of households suited to VLC DRWH usage.



As foreseen, tank size was a contentious issue, all users and most commentators favouring large tanks despite their apparent unaffordability: the nature of time-pressed research made it difficult to enforce price realism. Therefore initial concentration on very small tanks (capacities <1000 litres) was relaxed as the programme progressed.


Storage size strongly influences economic viability. Generally payback time increases and net present benefit falls with increase in tank size. This suggests a DRWH application ‘ladder’ progressing in steps from low investment (with short payback) to high investment (with high water security). The steps are: (1) opportunist collection of RW, (2) ‘wet-season’ DRWH or (3) ‘potable fraction only’ DRWH, (4) ‘all wet season water + core dry season water’ (‘adaptive’) DRWH, (5) ‘main source’ DRWH and (6) ‘sole source’ DRWH. This suggests DRWH implementation in stages; however the diseconomies of scale unfortunately penalise piecemeal installation of storage capacity. Moreover only under the most favourable relative circumstances should ‘sole source’ DRWH be considered: its economic returns are usually inferior to some degree of mixed sourcing of water supply, as is already the norm in most rural areas of LDCs. In terms of obtaining a good balance between economy and water security, Options 2 & 3 above look the most attractive in areas of Monsoon climate, Option 4 where dry seasons last 3 to 5 months, Option 5 in areas of bimodal or fairly uniform rainfall. The climate in the 8 communities studied generally suited use of Option 4, ‘adaptive DRWH’, where the payback time for a medium sized tank averaged 14 months using conventional tank designs, 8 months using the best of the new designs. This payback is based mainly on time-savings valued at half the local unskilled labour rates.


It was found that the ‘ladder’ concept could be widened to cover both tank sizing (and hence water performance) and construction quality, both together determining system cost. Sensitivity of tank cost to capacity averaged 0.55, so cost comparisons between rival designs can only be made ‘size-for-size’ or after converting to a ‘standard’ size such as 1000 litres. On this latter basis it was shown possible to lower system costs some 50% below previous best practice, e.g. to about $30 under Ugandan costs, by lowering apparent (but not significant) quality and by other measures. Some low-cost designs, like DRWH itself, clearly require familiarity before they become acceptable. Four out of eight new designs were considered worth recommending at the end of the programme. No clear advantage was found for factory as opposed to ‘on-site’ tank construction, despite claims for the former in the literature. Tank construction details are being posted on the web and are summarised in the Handbook.

Guttering was examined separately and provided they can be accurately aligned gutter sizes noticeably smaller than current practice are recommended for tropical climates – typically under 80mm width in a housing context. An optimum profile was found to be ½% slope over 2/3 of a gutter’s length, 1% for the rest.

Water Quality

Bacterial water quality, as measured by over 3000 E.Coli counts for tanks devoid of any entry filtration provision, was found to be highly variable site-to site. The RWH water had a median plate count of ca 30 FC per 100 ml, corresponding to the WHO ‘medium risk’ category. This is not very satisfactory and indicates the need to improve tank performance either by changes in design or in operation. RW water however had lower counts than those from the other sources used at 7 of the 8 sites (but not lower than in peri-urban Addis Ababa where the alternative standpipe water was of high standard). FC levels fall approximately exponentially with time after any rainfall event. Time constant T 90, i.e. storage time to improve by 90% (one WHO class), is around 3.5 days. Turbidity measures (NTU) showed a similar pattern of rising with rain events and falling quickly after them. Absolute turbidity measures would be significantly reduced by measures like first-flush diversion not used in these experiments.

Socio Gender

Interest in permanent DRWH was found to be high but the absence of affordable systems has been a constraint on their widespread use to date. At the household level there is a dilemma between tank capacity, durability and quality of water on the one hand, and cost considerations on the other. The fieldwork suggests there would appear a minimum size for a tank (of around 2000 litres) in order for DRWH to make significant impact on a household’s water fetching behaviour. To reap tangible benefits (in the form of substantial time or monetary savings) tank water must make a significant contribution to dry season water supplies. A smaller tank is sufficient in the rainy season to provide access to a convenient source of water which may also be of better quality than the traditional alternatives. Very low cost, small tanks (such as the tube of 800 litres) may have a special niche in emergency situations where time is of the essence, rather than long-term storage capacity or durability.

Poverty coupled with a lack of opportunity for the remunerative use of time saved may make people unable to invest in DRWH. Although they are concerned about the time they spend collecting water from traditional sources unless they can generate cash from the time saved, they have insufficient money to invest in RHW because there are too many other claims on the family purse. Poorer households, female-headed households and tenants often face additional barriers to adoption due to absence of property ownership, restricted space and limited access to credit and lack of information.

Project Reports


Inception Report

(April 2001)

(PDF Document) Full Document (44pp, 400kb)

Very Low Cost Domestic Roofwater Harvesting in the Humid Tropics:Existing Practice

(Jan 2002)

(PDF Document) Full Document (106pp, 900kb)

Very Low Cost Domestic Roofwater Harvesting in the Humid Tropics:Constraints and Problems

(Jan 2002)

(PDF Document) Full Document (38pp, 650kb)

New Technology for Very-Low-Cost Roofwater Harvesting

(April 2002)

(PDF Document) Full Document (25pp, 570kb) 

Very-Low-Cost Domestic Roofwater Harvesting in the Humid Tropics: User Trials

(June 2003)

(PDF Document) Full Document (47pp, 600kb)

Very-Low-Cost Domestic Roofwater Harvesting in the Humid Tropics: Its Role in Water Policy

(April 3003)

 (PDF Document) Full Document (14pp,340kb)