Structures Suitable for Emergency Storage in Tropical Countries (NRI) Part I: Survey and analysis Introduction Project background Objectives Method Results Assessment Critical factors in system selection Discussion Conclusion and recommendations

Structures Suitable for Emergency Storage in Tropical Countries (NRI)

Part I: Survey and analysis

Introduction

In all countries the system which distributes food from producers to consumers is planned on the basis of 'normal' conditions. Typically cereals not retained by farmers for their own consumption are collected through a network of procurement centres and moved to towns and cities, or for export; imports of cereals may be distributed in the cities. The system always includes storage facilities, to buffer the distribution process, as well as to accommodate seasonal cropping.

Abnormal conditions may lead to food shortage or surpluses. Shortages may arise suddenly, as a result of disasters such as earthquakes or hurricanes, or with some warning, in the case of crop failures, for example through drought. In emergencies the vital need is to get supplies to beneficiaries quickly. In addition to logistic problems, the climate is often hostile. High temperatures and relative humidities can accompany torrential rainstorms and gale-force winds. Any exposed foodstuffs will be wetted and will deteriorate rapidly if left undried. Any vulnerable structure will be severely damaged by storms or winds (Eaton, 1980). Therefore storage is a vital part of the distribution system which can provide food security especially when transport fails. Already-existing storage space in schools, army barracks and the like (FAO, 1983) can be used for transit storage in an emergency; but such temporary stores must be secure, because pilferage is a problem, and conveniently sited. Unfortunately, disasters can occur in inaccessible areas with little or no existing storage space.

Disasters may strike urban or rural areas, and existing infrastructure including food storage facilities may be inoperable. In the initial phase, food distribution is part of a general relief operation, and limited storage capacity is required very rapidly. Subsequent rehabilitation may (perhaps especially in rural areas) require new stores to permit food distribution for an extended period, although there may be no long-term requirement for these facilities. Similar considerations apply to the feeding of refugee groups.

Crop failures affect rural areas directly and urban areas indirectly. The existing system can probably distribute imported (donated) cereals to the normal urban population. It is, however, common in these circumstances for large numbers of the rural population to move to the towns, and additional emergency facilities may be needed to enable food to be distributed to them (UNCDF, 1985). To supply the rural areas it is often not sufficient to operate the normal grain procurement chain in reverse; storage may be in the wrong place, of the wrong kind, or too small to handle the quantity of food required by the rural population (UNCDF, 1985). Consequently there is a need for emergency food stores for a limited period until the next good crop is harvested. The supply of food aid to development projects may lead to similar physical requirements for storage facilities, but the time for planning and installation is greater.

A sudden food surplus resulting from an unusually good crop also gives problems. Once the national storage capacity is filled, there is a risk that the remaining harvested grain will suffer high losses. Emergency storage facilities should enable post-harvest losses to be kept to economically acceptable levels during the time the grain is held before entering the normal logistic system. Such storage is therefore distinct from food relief and will be treated separately in this study.

In Africa in 1985 relief food needs stood at 7 million tonnes; 24 countries were affected of which 15 have recurrently acute food shortages (see Table 1). The countries total 31% of developing Africa's population; nearly all have low incomes, negative changes in food production with high and continuing cereal food needs per capita. In parts of Latin America the situation is similar. On the other hand, in 1985 Burma and Indonesia had bumper harvests. (FAO, 1986a).

Table 1 Cereal food aid needs (1980-82). Annual changes in food production (1983) and GNP (1982) all per capita for African countries recurrently affected by food shortages (1977-82)

In 1986 there was adequate rainfall in most of Africa; 12 countries produced exceptional surpluses or bumper crops of cereals. At the time, Angola, Botswana, Cape Verde, Ethiopia, Mozambique and Sudan needed high levels of food aid. The total for Africa was 3.2 million tonnes (FAO, 1986a). In Asia in one country alone - Pakistan - the World Food Programme (WFP) were distributing food for 2.2 million refugees (Hauser, 1986) and in Bangladesh WFP had an even larger relief programme (FAO, 1 986a).

These data suggest that the need for high volumes of food aid inputs into Africa, Latin America and Asia will continue. In combination with erratic occurrences of bumper crops, this will cause continuing severe logistical and storage problems. For successful future food relief operations FAO suggest three guidelines.

A Improve early warning systems.
B Get food to the needy more quickly.
C Ensure that emergency food supplies are always available (FAO, 1985).

In this bulletin our concern is primarily with B, where emergency storage is an essential element in the race to distribute relief food without delay.

Logistical problems of the most critical commodities-bagged cereals- include their passage from often congested ports to main distribution centres or rail heads where they face bottlenecks; unloading ships can be delayed, warehouses are often already full and open-air storage with its associated losses is unavoidable. In this bulletin we describe how the appropriate emergency storage can be supplied to reduce stock losses.

Problems up-country are usually even more severe, with poor roads and unreliable transport. Correct choice of store type is therefore vital if facilities are to be installed rapidly and without excessive demands on resources. This study aims to make these choices easier. It does not consider questions of strategic storage which arise in connection with FAO guideline C.

The Overseas Development Administration (ODA) is directly involved in emergencies through funding, through its Disaster and Refugee Unit, and through the Overseas Development Natural Resources Institute (ODNRI). ODNRI advises WFP, other multilateral agencies and individual countries on food loss reduction, in particular on grain store management and construction.

Project background

ODNRI has been asked by WFP and other agencies for practical advice on emergency storage operations, frequently in circumstances where the need for storage has not been recognised until after the relief food has arrived in the recipient country. In these circumstances the only option open may be to airfreight tarpaulins or temporary structures as near to the site as possible. ODNRI has contributed to the continual process of redesign and modification of one type of structure which has been used in this way. Cough (1979) carried out exposure trials of various types of sheeting used for these structures.

In 1980 reported failure of flexible silos for emergency use in the tropics prompted ODNRI to conduct a field survey (O'Dowd and Kenneford, 1982). A team of engineers visited seven tropical countries and one of their recommendations was that donor agencies should investigate alternative systems for bagged stock. The present study was commissioned by ODA to evaluate different systems and structures suitable for emergency storage and currently available.

Objectives

The aim of the study is to make the choice of emergency storage system easier for donors and governments. To achieve this there are three objectives: (a) to provide donors with a technical, managerial and economic assessment of storage systems currently used for emergencies, (b) to identify critical factors in system selection and lastly (c)to suggest new systems suitable for testing overseas while providing manufacturers with guidelines on design and materials.

For donors' convenience we summarize these findings in a users' guide, including a list of suppliers, in Part II of the bulletin.

Method

An ODNRI engineer visited five food relief sites in Botswana in 1984 to advise on construction of stores for emergency supplies. He visited again in 1985 to report on the condition and performance of these stores as well as other emergency storage systems in operation (UNCDF, 1985; O'Dowd, 1986). To this assessment was added that of the ODNRI officer assigned to manage the food relief operation following Hurricane Isaac in Tonga (Morley, 1987) where similar warehouses were erected.

To achieve wide coverage of emergency store users, ODNRI enlisted the assistance of professional relief workers by contacting the WFP, the Catholic Fund for Overseas Development, Christian Aid, Oxfam, Save the Children Fund and the Tear Fund, obtaining names and addresses of workers with experience of emergencies and writing a letter to each (see Appendix 1) asking for their assessment of field performance and qualities essential for such structures. ODNRI officers supplied details of food relief in Ecuador (Calverley, 1987) and Nicaragua.

To these reports were added ODNRI officers' recent notes of emergency storage for bumper crops in Burma, Indonesia, Liberia and Zambia. ODNRI engineers contacted 40 manufacturers for details of a wide variety of rapidly erected structures suitable for consideration. (see Appendices 2 and 3). The information was checked with each manufacturer and is provided in Part II of this bulletin.

Lastly, from all the available information, the most pressing design problem was selected to formulate a research proposal.

Results

The use of steel warehouses for emergencies

In the 1985 assessment in Botswana four criteria applied:

(1) do the stores function adequately or are additions required?
(2) were they built to schedule?
(3) were costs approximately as estimated? and
(4) was permanent storage justified or would temporary storage have sufficed?

The store specifications (see Appendix 4) were for locally manufactured 500-1000-tonne capacity portal-frame steel-clad warehouses, prefabricated for rapid erection and with the function of providing transit protection for relief food.

Storekeepers and depot managers stated that these new drive-through stores functioned well (see Plates 1 and 2); they were secure and easy to manage, and truck turn-around time was 65-75% faster than when using the old stores, using the same labour and transport as before. The roofs were designed for maximum wind loads as were the structures themselves. Despite severe storms, no complaints have been received regarding construction. No buildings are perfect on completion and in this operation the main omissions were site-roads, hard-standings, drains, and emergency storage plinths which had been specified. Besides site roads, hard-standings at all sites are essential in wet weather for easy truck turn-around. The urgency of this building project, shortage of trained staff and the remoteness of the sites meant that topographical surveys and searches were not made. Skimping site surveys caused extra costs later and one site choice had to be discarded; at another the site slope required large quantities of earth to be moved. These operations caused delays but once contractors were mobilized, store construction took less than four months on average.

These operations also, of course, caused capital costs which therefore exceeded estimates by approximately 9% and the above-mentioned omissions also had to be paid for. Actual storage costs/tonne were calculated from capital costs using the annual throughput, the useful life of the store and the rate of interest. The annual cost per tonne in 1985 varied from £4 at stores near the capital to £5 for remote stores.

To calculate financial viability, estimates of losses in temporary and permanent storage as well as costs of temporary storage were obtained.

Storekeepers and depot managers were unanimous: the combination of tarpaulin storage and inexperienced management resulted in severe losses. This was particularly true with vulnerable food like Corn-Soya-Milk (CSM) when there was a lot of damage to sacks and tarpaulins caused by baboons and rats, which in turn caused considerable spillage (O'Dowd, 1986). But it was in the wet season when there was flooding and moulds and insects were most active that heaviest losses occurred. With these increased losses and without good management, temporary storage was more costly than permanent storage (using annual costs). With good management, temporary storage was cheaper unless it was stipulated that permanent stores had a useful life of from 4 to 7 years combined with a throughput ratio greater than 6:1. (The throughput ratio is the quantity of produce passing through the store each year divided by the store capacity).

In theory, storage costs/losses were minimized when permanent stores were used for vulnerable foods and tarpaulins used for durables. In practice, managers stated that this is difficult to implement because supplies arrive on site in random order. Storekeepers and depot managers considered that transport was the most critical factor in store operations, and in 1985 every tonne of food at remote locations bore £8 transport cost.

In this relief operation the role of government had been vital. Only by involvement at the highest level could the government harness resources, predict emergencies in good time and instil a sense of urgency to provide adequate relief.

In Tonga, following Hurricane Isaac in 1982, tarpaulins were initially used for emergency storage, followed by the importation of prefabricated steel-frame, steel-clad warehouses not dissimilar to those manufactured in Botswana (Morley, 1987). The stores were donated by ODA but supplied by a New Zealand firm because this was faster and because standard buildings in New Zealand are designed for high wind loads. Storage was satisfactory using low-cost, low-weight, polyethylene-coated, woven polyethylene tarpaulins until the prefabricated buildings were erected. The building was used with a temporary coral rock floor until a concrete floor could be laid six months later. An important point which arose out of the Tongan experience was that the combination of skilled management and labour was essential to work tarpaulins and permanent storage satisfactorily. When management and labour were unmotivated and untrained, damage to tarpaulins, warehouses and stock resulted.

Reports by relief workers

In Asia and South America relief workers reported the use of local materials for emergency storage. Calverley (1976) inspected local store construction in Ecuador, where timber framework supported plastic cladding to provide 250-tonne capacity 'green-house' type structures (see Plates 3-4). Morton (1987) describes how in Thailand local manufacturers provide similar large frame tents to WFP specifications. These structures were easy to transport by lorry, could be relocated as necessary and could also be used as mobile clinics and supplementary feeding centres. Relocation was a very important operation and function. In Pakistan, Hauser (1986) reports on earth-built stores being upgraded to locally made brick stores which were part of the relief food distribution chain. With the exception of southern Africa (where, for example, Botswana used local materials) in the rest of the continent and especially in north-east Africa, relief workers reported that local materials are rarely used for emergency storage, although locally built warehouses were planned in Sudan (O'Shea, 1984). Eucalyptus poles are sometimes used for frames and for dunnage but usually this item had to be made up from imported timber shipped to the port of Assab (Morton, 1 987).

With regard to imported emergency structures, all relief workers stated that the highest priority was fast easy erection by local staff to provide shelter quickly. Ease of transport and relocation were also important. In this context, lightweight tarpaulins or plastic sheeting were a good first-aid measure (Reece, 1987) and essential, combined with other store types, to the success of every emergency operation. Initially, tarpaulins protected the first relief food to arrive at a crowded port. One donor actually shipped tarpaulins with relief food, but normally tarpaulins are supplied by air freight. Tarpaulins with eyelets and ropes were particularly useful because these could be tied down and even nailed and were worth the extra cost over plain plastic sheeting (see Plate 5). Tarpaulins are flexible and easily transported/relocated (Morton, 1987) but do not provide the shelter necessary for mixing food rations or the facility of easy stock management, especially in wet weather. Timpson (1987) speaking from seven years' experience of relief food operations in Africa states:

'....the ideal store should not require expatriate assistance for assembly and erection. It should be light and manoeuvrable enough with the minimum of parts so that a group of relatively unskilled local staff can erect it in as short a time as possible but certainly within four days....'

Other relief workers echo this reluctance to use expatriate assistance, but Timpson goes on to say that in 1980 relief agencies in Uganda purchased a number of mansard-shaped, steel-frame, plastic-clad stores which were erected in three days with the firm's instructor helping initially. These structures proved robust, any tears were easily repaired and the stores have lasted until the present time (1987). Several stores were relocated without problems. More of the same stores (300 tonnes) were purchased from Eastern Sudan. One criticism of the stores was their vulnerability to pilferage; people will cut through the plastic cladding and remove food. For examples of this type of store and its erection see Plates 6-11.

Another relief agency purchased 500-1000 tonnes stores of a different design. With the services of two engineers each larger store took 2-3 weeks to erect Later one store was completely destroyed by a violent storm (suffering wind damage) while the 300-tonne stores were undamaged. The dangers of high winds are emphasised by Winer (1987) who reported that structures are more vulnerable to strong winds when flaps and ventilators are open (for essential ventilation). In Mali, Hodges (1987) reported severe wind damage to two plastic-clad stores (see Plate 12) designed for wind speeds of 46 metres per second where winds of 67 metres per second prevailed*. In this instance the store doors would not close, leaving a gap which may have allowed wind to enter and, because the manufacturer's instructions were ignored, the structure foundations consisted only of steel pins in the sand. These errors may have combined to cause failure: Fortman (1987) reports that in 1985 no fewer than 15 aluminium-framed buildings were destroyed by winds in the Sudan.

In the United Kingdom, tents are known to be vulnerable to wind damage (Houghton and Carruthers, 1976). Overseas, several workers reported that tents were successful. These were quick to erect and easy to relocate. In Eritrea the satisfactory use of tents was reported by Hill (1987), who mentioned that tents lasted up to five years and there was no security problem, while other workers drew attention to the vulnerability of tents to pilferage in some countries. Plastic-clad stores, tents and tarpaulins all require dunnage in the wet season to raise food sacks above ground. Rain can also damage steel-framed structures. We recently saw a relocated store in Ethiopia where the cover had been inadequately fixed to the ground; this caused it to sag and lodge rainwater which in turn caused partial collapse.

Prefabricated, curved-steel-clad, frameless, modular stores (see Plate 13) functioned satisfactorily, were secure and were more durable than plastic-clad types but were reported to be slower to erect. Timpson (1987) relates that these frameless structures often took two weeks or more to erect and therefore frustrated plans for pre-positioning storage. It is reported that frameless structures require a higher degree of skill to erect than plastic-clad structures. Foundation work must be to a high standard (Winer, 1987) and in one instance an engineer took some three weeks to erect the first structure and then a further three weeks to erect two more. Rees (1987) in the Sudan saved the firm's erector's fee (£4000 with all items) by erecting one of the structures in two weeks and Morton (1987) refers to 'fairly quick' erection over a concrete plinth.

It was stated that frameless structures were more difficult to move from the port to the rural area than plastic-clad structures. In addition, plastic is an easier material to repair. Frameless steel structures in Ethiopia were corroded and will need maintenance to last the 15 years that the manufacturers guaranteed (Timpson, 1987). That frameless steel structures are damaged and rust in transit is confirmed by Winer (1987). Morton (1987) and Rees (1987) both considered these structures functioned satisfactorily for the period of the emergency but they were criticized because they were somewhat inadequate for long-term storage of food commodities and unsuitable for alternative uses. It should be noted that these reports are based on a limited response to our wider survey while the manufacturers, Conport Structures Ltd. report that they have commissioned 150 modular buildings in Ethiopia and these have had a totally favourable response. Large stores of 1000-tonne capacity were particularly unsuitable after the initial emergency phase when there is need for longer term rehabilitation assistance (see below). Frameless structures were useful at the beginning and end of the major supply lines, for example, between Port Sudan and Nyala, but required fork-lift trucks to fill to capacity, either because unskilled workers could only stack 10 plastic sacks high, or the sacks split at the bottom if built more than ten high, because they were poorly stitched by piece rate labour in the port (Walker, 1987). In subsequent calculations we assume that bag stocks are 2 metres high. This limitation increases the capital cost per tonne of warehouses.

The range of imported and locally built stores available

There is a wide range of emergency stores available, both imported and local, bag and bulk. When classified in terms of cladding, frame and foundations, they fall into five types (see also Table 2.):

A Structures with rigid frame, rigid cladding and concrete foundations - typically traditional warehouses (Type A1), and Nissen huts (Type A6) for bag storage. See Plate 14.
B Structures usually without a frame but often with pre-stressed modular panels and concrete foundations - typically silos (B4) and similar frameless structures for bulk grain (Type B3). See Plate 13.
C Structures with flexible cladding, a variety of lightweight frames and ground anchors - typically rapidly erected bag stores (Type C1). See Plates 6-11.
D Structures with flexible cladding, rudimentary frames and a variety of foundations - typically tents (Type D1) or tarpaulins for bag stack storage (Type D3). See Plate 5.
E Quasi-structures without frame or foundations like bunkers (Type E3), clamps (Type E2), pyramids (Type E1) and pits (Type E2) for bulk grain storage.

Table 2: Available emergency store types: local and imported, with examples of both bag and bulk

As a general rule, bag stores (Types A, C and D) are more often used for food relief while bumper crops are often stored in bulk (Types B and E). There are exceptions - for example, structures of the frameless, steel-clad (Type B3) have been used for both bag and bulk storage.

Each type discussed above is part of a storage system composed of a store, handling methods and transport, all controlled by management to achieve a stated function. Thus in Southern Africa drive-through bag stores (Type A1) are part of a rapid-transit storage system involving the intensive use of scarce transport vehicles and limited numbers of store labourers (UNCDF, 1985) (see plates 1 and 2). On the other hand, cover and plinth (CAP) bag stores (Type D2) are part of a seasonal overflow storage system which demands high levels of on-site management and intensive labour input (Gary, 1985). The system has been adapted in India for storage of surplus production at buying centres (see Appendix 5). These two examples give some clue to the distinct natures of food relief and bumper crop emergencies. For instance, drive-through stores are filled and emptied 6-9 times a year, while CAP stores are usually constructed and emptied once in a season. Further examples of emergency storage systems are given in Appendix 6 and further information on storage structure sub-categories is given in Appendix 7. A comprehensive computer literature search was carried out by ODNRI library but revealed no published data on bag or bulk stores for food relief/bumper crops. Data were obtained from relief agency, WFP and ODNRI sources, including bumper crop storage in the next section.

Bumper crops

Bumper crops need extra storage space initially near the production site, and/or at buying centres later (O'Dowd and Kenneford, 1982). In Burma, bulk paddy is stored on the ground out of doors, in pyramid-shaped heaps sloped at the angle of repose (see Plate 15). Security is achieved by spreading paddy husk ash around the pyramid perimeter. If the grain is disturbed in any way, it will roll down, cover the ash and signal pilferage. The bulk is moved by basket into temporary stores as the rains approach. These stores are constructed of locally obtained wooden frames covered with woven fibre matting (see Plate 16). This system is subject to losses (Tyler, 1987). A more commercial version employs a concrete floor and a perimeter wall. Grain is heaped on the floor mechanically as high as its angle of repose will allow, covered with a PVC sheet which is anchored and sealed to the wall. Hermetic conditions can be achieved and the system has been tested by CSIRO and is used for overflow storage in Australia and the United States (Murray, 1987).

For bagged grain (see above), Garg (1985) described the development of CAP storage in India over 30 years. CAP storage is designed for bumper or surplus storage and, it is claimed, costs about one-tenth of the initial capital cost of warehousing. But CAP storage requires very careful management (see Appendix 5).

Flexible silos have been used successfully for bumper crop storage (Kenneford and O'Dowd, 1981), but are unsuitable for food relief because frequent opening and closing of the silos causes wear and tear. Like CAP storage, flexible silos require careful management. The silos are suitable for bag or bulk storage, the standard size being 500-tonne capacity. The silos are supplied with rodent guards and currently use a cup-shaped liner of PVC-coated polyester fabric. This liner is supported by a circular welded mesh frame composed of bolted sections. Bags are loaded into the cup until level with the mesh and then into a cone. A conical cover fits over the cone, is rolled into the cup liner and tied down to form a waterproof and hermetic container suitable for fumigation. For full details see Part II.

Emergency overflow storage is also possible in steel silos. In Liberia the overflow from the central buying point was stored in imported steel silos. Bulk storage has the advantage that no bags are required, but the disadvantage that silos require to be aerated to reduce the risk of moisture migration for safe storage exceeding one month (Cough, 1987).

The choice between bag and bulk may depend on a number of factors including which type of transport a country has adopted.

Assessment

Relief food stores

Locally built warehouses

Locally built warehouses made financial sense where throughput was high (1:6), where the emergency was longer than four years and where there were existing temporary storage facilities. Technically and from the management standpoint, warehouses are demonstrably superior to tarpaulin storage in Botswana (O'Dowd, 1986). But for a short-lived emergency with no existing storage the situation is different. O'Shea (1984) suggests that for local store building to succeed (in Sudan) prerequisites are: swift decision-taking, a careful selection of suitable contractors, availability of diesel fuel and proper contract supervision. In a short-lived emergency these are the very factors in short supply. Even in Botswana when food relief stores were built to high standards, unavoidable delays and cost over-runs caused omission of items vital to efficient operations. Unless store building can receive continual skilled management input, monitoring progress and quality, locally built warehouses are not an option; if there is any doubt reliance can be placed on well-tried imported emergency storage structures.

Imported steel warehouses

The warehouses imported into Tonga functioned satisfactorily but were slow to procure and erect, even though obtained by the shortest method from New Zealand. As shall be shown, a standard warehouse specially designed for easy freight and rapid erection (see Appendix 8), cannot, allowing for all delays, be procured in less than four months. The size of warehouse chosen should be influenced by its likely use post-emergency. In general, two buildings of 500-tonne capacity are likely to be more valuable locally than one 1000-tonne capacity building, for storage, factory work, community hall, etc.

Imported 'temporary' stores

When urgency is the keynote, procurement of traditional warehouses is usually too slow, the warehouses need too much time to erect and also require scarce and costly expertise. Even a specially designed warehouse is at least 50% heavier than a plastic-clad store and is therefore difficult to relocate (see below). Given that traditional warehouses are slow to erect and difficult to relocate, the alternative choices for emergencies which have been tested and tried overseas are limited- Relief workers found that frameless structures functioned well and were secure, but were relatively slow to erect and, it was reported, too difficult to relocate, because of their weight and because level strip foundations are needed at the new site. Besides that type of structure, only plastic-clad stores, tents and tarpaulins have been tested overseas, all of which require dunnage or ground sheets. Plastic-clad structures were quick to erect and straightforward to relocate. Stock management of bag stacks in such stores was not a problem. These structures are not, however, secure from theft, and certain designs with inadequate foundations were destroyed by wind. Similarly, tents were quick to erect and easy to relocate; stock management was adequate but security was poor. Tents are also vulnerable to wind damage. Tarpaulins were very flexible but presented more stock management problems than plastic-clad stores or tents especially in wet weather. For comparative purposes these operational factors are contrasted for each type of store in Table 3 (see p.13). Operationally no one store is superior in their qualitative assessment, but where ease of erection and ease of relocation have priority, tarpaulins, tents and plastic-clad stores in combination or singly are preferable, according to reports from relief workers, to frameless (steel) or portal-frame stores.

Table 3: Relief workers' reports of operational factors in store choice

Table 4: Capital costs, procurement periods and packed weights of imported structures for food relief

This qualitative view is reinforced by quantitative data. Using a stacking height of only 2 metres, capital costs/tonne, procurement periods and packed weights are compared for each store type (see Table 4, p.13). Tarpaulins and tents have lower capital costs, are faster to obtain and erect, and are one-tenth the packed weight of other store types. Plastic-clad stores are fast to erect on site and are often lighter in weight than the example shown in Table 4, especially when the capacity is smaller. For the two years' length of most emergencies, annual costs of tarpaulins and tents are also lower than for other stores (see Table 5). The effects of longer store life are discussed below. The data for Tables 4 and 5 are derived from Appendices 9 and 10 respectively.

Table 5: Effect of store life on annual costs per tonne (£/t)

Bumper crop storage

The main types of bumper crop store have been described in the section Bumper crops. The main distinctions from relief food storage are that farm-gate prices for bumper crops are far below the on-site cost of relief food (FAO, 1 986a). The throughput ratio is the quantity of produce (tonnes) passing through the store each year divided by the store capacity (tonnes); many relief food stores have a throughput ratio of 9:1 (King, 1987) while bumper stores usually have a throughput ratio of 1:1. Assessment will be possible with on-site investigations of operation, flows, loss levels and costs. We have attempted to arrange suitable visits without success, but hope to in the future.

Critical factors in system selection

On a relief food site, typically subjected to high winds, rains and to pilferage, four main factors influence system selection (imported or local) in the following order:

- how much warning was there?
- what level of funding is available?
- what level of site management prevails?
- what local transport and available store labour is there?

As mentioned in the introduction, FAO (1985) emphasize the need for early warning; in practice there is seldom much time left to instal a storage system; funds are usually scarce although relief agencies praise the generosity of donors (Reece, 1987).

Management is all-important for efficient operation and relocation of stores and also for coping with unforeseen problems (O'Dowd and Kenneford, 1982). Transport is often the limiting factor, together with available labour in relief food operations, (UNCDF, 1985).

In Part II of this bulletin these factors are taken into account using a decision tree for users' guidance (see p.21). One of the many options available-airwarehouses-requires special mention. ODNRl's experience is that airwarehouses require a higher level of day-to-day management because they are dependent on continuous operation of an electrically powered fan. If the store is deflated it must be protected against wind damage. Wind can be a problem for all structures and in selecting a storage system donors must check that the store has the correct design wind speed. Many manufacturers consider that a design wind speed of 47 metres per second is sufficient. Eaton (1980) states that while mean wind speeds in tropical storms may exceed 33 metres per second the gust speeds are far greater locally and the design wind speed for Mopti in Mali, based on 12 years data, is 72 metres per second (Meteorological Office, 1987). Donors must also ensure that manufacturer's instructions for adequate foundations are adhered to, especially when the soil is sandy. Therefore when preparing tenders for emergency structures, donors should include the following questions:

A What is the structure design wind speed?
B What form of foundation is supplied for use on sandy soils?
C Is the structure designed for tropical exposure? Other points mentioned earlier include:
D How many structures are available ex stock?
E Is erection without a supervisor practical? Are visual instructions/local-language manuals available?
F How long should erection take?
G Is the structure easily relocatable?

These points are summarised in Part II of this report.

An additional critical factor in system selection is the need for cooperation between the donor and structure supplier to achieve the objective of getting food to the needy quickly. Donors and suppliers must be flexible to take account quickly of congested ports and overcome obstacles and delays, for example by using airfreight for a portion of the order. At the recipient country the role of government is critical, for example, in allowing freight to be cleared duty free without delay.

Discussion

For relief food the priority is to provide shelter quickly. FAO (1986b) plan to pre-position stocks of relief food; could emergency structures for these stocks be included? For the moment we might approach ODA Relief and Disaster Unit and some British manufacturers to suggest that a minimum number of tarpaulins, tents and plastic-clad stores -which relief workers use and approve of- be held in stock in readiness.

Tarpaulins, tents and plastic-clad stores are easy to handle/ repair, quick to erect and easy to relocate with the very slender resources available. Managerial, technical and financial evidence supports this view (see Tables 3, 4 and 5). However a pressing need is to help United Kingdom manufacturers of plastic-clad stores (see pare. 3, Objectives) to design for tropical wind loads, which are often double United Kingdom loads. Farm Building Division, AFRC Institute of Engineering Research, are investigating the effect of wind on plastic-clad structures and are keen to collaborate with

ODNRI as are the Overseas Development Research Unit of the Building Research Establishment who have experience of tropical conditions.

We first investigated the use of steel-clad portal frame warehouses for relief food and concluded that warehouses have certain advantages over more temporary stores, like plastic-clad structures. Over six years' life costs and benefits can be compared, from Table 5:

	£
Steel-clad portal frame warehouse, annual cost/tonne	1.85
Plastic-clad steel frame warehouse, annual cost/tonne	1.36
difference:	£0.49

If labour and maintenance costs are equal, the steel-clad portal frame structure is justified if losses are reduced by only 1 kilogram per tonne of relief food priced at £500 on site. Such reductions in losses have been recorded where steel portal-frame warehouses with smooth concrete floors replace temporary structures on hardcore floors. Spillage from burst bags can be recovered from concrete floors and wear and tear on sacks is less (UNCDF, 1985, O'Dowd, 1986) Management is also easier.

Supporting the provision of permanent locally built warehouses for emergencies, O'Shea (1984) reporting to the Sudan Government, stated: 'the problems of storage in the long term have been somewhat overlooked in the desperate need to relieve the drought victims'.

In theory, if early warnings were the rule, locally built steel warehouses would be selected. In practice, warnings are late (FAO, 1986a) and relief agencies and governments are fully stretched to beat the clock. Initially, at least, there are no opportunities and no resources for long-term planning. When the worst of the emergency is over, some form of collaboration between relief agencies and governments may be possible.

In some countries there may be an opportunity for permanent and temporary (emergency) storage to complement each other in improving relief food distribution to famine-affected areas. In Ethiopia permanent storage for 'normal' relief food flows will be developed initially, but it may be uneconomic or impractical to provide permanent storage at all levels for the additional irregular grain flows occuring during a major emergency such as the 1984/85 famine. Instead, preparations should be made to provide at short notice safe and economic emergency storage at strategic locations for these additional and irregular flows.

Finally, there is some difficulty in arranging visits to bumper crop sites to see the intake and storage operation because notice is short, but we hope to overcome this problem and to obtain information comparable to that obtained for food relief.

Conclusion and recommendations

From this investigation we conclude the following:

(a) There is no pressing requirement for new design of emergency storage buildings, but there is an urgent and over-riding need for re-designing existing patterns of fabric/sheet-clad structures to ensure safety from tropical wind loads.
(b) There are insufficient data available on 'on-site' studies of bumper storage systems and a lack of reliable measurements or estimates of losses to support claims that losses are either unacceptably high or acceptably low.
(c) The appropriate choice of an emergency storage system for food relief can be readily obtained by use of the critical factors in system selection in conjunction with operational and cost considerations. These are summarized in Part II of this bulletin.

We recommend the following:

(a) Donors should select an emergency storage system for relief food heeding critical factors in system selection; by use of the decision tree shown in Part II, users can combine managerial, technical and financial assessments shown in Tables 3, 4 and 5, and summarized in Part II. The list of emergency stores suppliers is included.
(b) Donors should collaborate with appropriate firms to ensure that a minimum level of tarpaulins, tents and fabric-clad structures are held in stock in readiness for an emergency, and should study the feasibility of funding, in whole or in part, such a stock.
(c) A research and development project should be undertaken in collaboration with industry to redesign and cost selected emergency storage structures so as to obtain units that are suitable for high wind loads.
(d) A sequel should be undertaken to the study here reported to:

(i) up-date the information presented; and
(ii) obtain 'on-site' data of bumper crop storage and the losses occurring in such systems so that a full evaluation of the various structures available can be undertaken.