Agri

IEE REVIEW

Electronics in UK agriculture and horticulture

S.W. R. Cox, QBE, BSc, CPhys, FlnstP, FIAgrE

Indexing terms: Instrumentation and measuring science, Control equipment and applications, Measurement and measuring

Abstract: This review traces the developing con-tribution of electronic instrumentation andcontrol systems to UK agricultural and horticul-tural production. It covers applications in thelivestock, arable and horticultural sectors in turn,taking the important grass crop as part of thearable sector. These main sections are subdividedby type of livestock, crop or farming operation, asappropriate. The section on livestock dealsseparately with feeding, weighing, quality assess-ment and environmental control in poultry, pig,sheep and cattle production. Much of the materialon cattle is concerned with developments in dairyparlour automation, including automatic identifi-cation of cows. The section on arable crops andgrass covers instrumentation and control of trac-tors and implements generally before dealingseparately with planting, spray and fertiliser appli-cation, harvesting, crop drying and crop gradingand storage. In the section on horticulture, fruitand vegetables are considered separately fromprotected cropping in greenhouses and mushroomunits. Forecasts are made of likely commercialdevelopments in each of the three sectors. A con-cluding section provides an overall assessment ofthe present status of electronic monitoring andcontrol equipment in agriculture and horticulture,together with the objectives of research and devel-opment in this sphere.

1 Introduction

Until this century instrumentation in agriculture andhorticulture was mainly concerned with weights andmeasures for trading purposes. Temperature measure-ment found a place earlier than this — although on alimited scale — in greenhouses and dairies. Otherwise,farmers and growers everywhere relied upon traditionand experience to judge the state of their soils, crops andanimals before making the many, varied and oftencomplex operational and management decisions that areneeded for successful food production. Then, in the 1930s,instrumentation of field machinery began when tractormanufacturers followed the automobile industry in fittingelectrical gauges to their machines. At the same time, the

Paper 5371 A, received in final form 3rd September 1986. CommissionedIEE ReviewThe author was formerly with the National Institute of AgriculturalEngineering, Wrest Park, Silsoe, Bedford MK45 4HS, United Kingdom.He now resides at 18 Lister Avenue, Hitchin, Herts, SG4 9ES, UnitedKingdom

spread of farm electrification [1] led to the widening useof electrical instrumentation and control equipment forheating, ventilation and illumination of animal houses,crop stores and greenhouses.

Electronic equipment began to appear on the farm inthe late 1940s and early 1950s, making its first majorimpact by facilitating the measurement of moisture incereals during and after harvest. Since then, farm elec-tronics has slowly but steadily extended into manyspheres of crop and animal production, greatly assisted inrecent years by the development of microelectronics. Thisdevelopment has considerably reduced initial concernabout the cost, complexity and reliability of electronicequipment under farm conditions. In the UK manyfanners and growers now accept that electronic instru-ments provide reliable data which they can use toimprove the short-term and long-term management oftheir enterprises. Many also accept the value of regular,automatic monitoring of particular operations. Fewer areyet ready to hand over their operations to automaticcontrol systems, because, in some circumstances, a mal-function or breakdown of the control equipment couldput a whole year's production at risk. Nevertheless, auto-matic control systems have been introduced successfullyon UK farms. In general, farm electronics is no longer anovelty and it will find application where it fits intofarming practice and can offer an economic return on theinvestment incurred.

A review of farm electronics must place emphasis onthe economic issues. Over the past 40 years manyattempts have been made to develop electronicequipment for farm use. The literature on this subject isextensive. Most of these developments have failed to gainacceptance in farming and the reasons have been moreoften economic than technical. The electronics engineerwho works with the agricultural engineering industrytherefore needs to appreciate the special problems thatare posed by farming, to judge how and where electronicsis likely to succeed commercially.

It is the purpose of this Review to describe electronicequipment which has found application in food pro-duction, together with the farming and economic factorswhich led to its uptake, and to indicate areas for furtherdevelopment, in the light of those factors. Although farmelectronics is established internationally — particularly inWestern Europe and North America — the Review con-centrates on progress in the UK. This does not reducethe range of topics covered appreciably, because farmersand growers in the UK have been among the first toadopt electronics, on a broad front. The factors whichhave influenced their attitude can be deduced from thedevelopment of UK food production over the past 40years.

466 IEE PROCEEDINGS, Vol. 134, Pt. A, No. 6, JUNE 1987

Post-Second World War, the UK has been — and stillis — committed to a large measure of self-sufficiency infood production, to meet the needs of a population thatnow exceeds 56 million. Self-sufficiency is currently over80% in the foodstuffs that can be produced under UKconditions. However, in the same period there has been asteady decline in the farming workforce to its presentlevel [2] of about 700000 farmers, managers andemployees (full- and part-time), which is a much smallerproportion of the national workforce than elsewhere inEurope or North America. This achievement has beenmade possible by intensive methods of production onhighly mechanised farms and horticultural enterprises ofincreasing size. Crop and animal breeding, coupled withthe use of a wide range of agrochemicals and the improv-ing products of the agricultural engineering industry havegreatly increased yields per hectare and per animal.However, these developments have also brought prob-lems such as environmental pollution caused by thewastes from intensively housed animals; drift of cropsprays; safety hazards to farm staff and livestock, and agreater dependence on fossil fuels (directly and indirectly).Public concern over many aspects of intensive farminghas to be reckoned with. In addition, the large food pro-cessors and retailers who now meet most of the needs ofthe home market are insisting on produce of ever higherand more uniform quality, delivered on schedule, to meetchanging consumer standards. When these factors areadded to increasingly intense international competitionin the marketing of food and food products, both athome and overseas, the UK producers have needed totake advantage of every available aid to improve effi-ciency and control over their operations to remain profit-able. It is therefore not surprising that innovativeproducers have been ready to adopt electronics (indeed,some have been pioneers in this field) and others havebeen quick to follow.

The nature of the market for farm electronics variesconsiderably from sector to sector in agriculture and hor-ticulture, therefore each sector will be examined separa-tely. Tables 1-3 provide some relevant background

Table 1: Agricultural use of land in the UK, 1984 [2]

Table 2: Quantity and value of UK farm produce, 1984[2-4]

Commodity Area/Number Yield, Mt Value, (£M)

Total land areaTotal agricultural areaAgricultural use:

young grasspermanent pasturerough grazingall other crops

AreaMha

2418.7

1.85.16.15.7

%

10078

7.52125.524

information on the relative sizes of the different sectors,together with the scale of UK agriculture and horticul-ture overall. However, before embarking on the sectorreviews, some general comments can be made about thismarket.

First, given the variation in the environmental condi-tions under which most food production takes place inthe UK, together with the variability of the crop andanjj^al products themselves and our limited models ofcrop and animal development, the establishment of reli-able methods of measurement and data analysis for farmuse is usually a slow and difficult process.

Second, the environment the equipment must with-stand is often harsh, yet farmers and growers expect ahigh degree of reliability from it, combined with ease of

Crops:cerealspotatoessugar beetother arable cropsall horticultural crops

Total crops

Livestock:cattlesheep and lambspigspoultry

Livestock products:milk and milk productseggsother

Total livestock sectorTotal crops and livestock

Area, Mha4.00.20.20.70.2

Numbers, *108

13.234.8

7.7127

26.57.39.0

——

Meat, Mt1.130.300.960.85

15590 Ml12300 M

2361582242356

12414782

1922579

1000664

2293537156

715111933

Table 3: Selected inputs — UK agriculture, 1984 [4]

Input Cost, £M

Feeding stuffsMachineryMaintenanceFertilisersSeedsLivestockOther*Total expenditureLabourt

2857943246963264180

106265151925

* Includes fuel and power, spray chemicals, veterinary expensesf Hired, family and partners

use and minimal maintenance — all at low cost. Fortu-nately, a high level of accuracy is rarely justified. Exceptwhere trading standards are concerned (i.e. at the market-ing stage) accuracies of ±2-3% are normally sufficient.Facilities for multiple sampling and averaging of results,combined with ready calibration and fault diagnosis, areof greater importance.

Third, the availability of suitable sensors is often amajor problem. Although some standard industrial trans-ducers have found applications on farms, many othersare too costly, frequently due to their high performance.In the latter cases suitable low-cost sensors are unlikelyto be manufactured for the small agricultural and horti-cultural market, which must hope to take advantage ofdevelopments for other industries. The manufacturer ofnew types of sensor specifically for agriculture or horti-culture faces this problem in greater measure. This is oneof the most serious barriers to the extension of farm elec-tronics.

Finally, little information can be given on design andperformance standards in farm electronics at present, dueto the often isolated and still formative nature of develop-ments in this field. Nevertheless, the British StandardsInstitution (BSI) and International Standards Organis-ation (ISO) Committees responsible for promoting stan-dards and codes of practice for farm buildings andmachinery have begon to address this subject and ameasure of standardisation can be expected to emerge inthe 1990s.

2 Livestock production

In the UK the livestock sector accounts for almost two-thirds of the total value of agricultural and horticulturalproduce. Within this sector, milk and milk productsmake the largest contribution (Table 2). The importance

IEE PROCEEDINGS, Vol. 134, Pt. A, No. 6, JUNE 1987 467

of the ruminant animals (cattle and sheep) is a conse-quence of the abundance of the country's main crop —grass — particularly in the west, with its higher rainfall.The need for cattle and sheep farmers to make best use ofthis valuable crop can be seen by reference to the cost ofanimal feedstuff's (Table 3). For all livestock farmers, theoptimal use of feedstuff's is a major concern.

Another feature common to the livestock sector is itsgreater use of labour, arising from the need for regularattention to the animals, in the interests of their welfareand safety. Despite this, automatic monitoring andcontrol systems have developed faster in the livestocksphere than in the arable sector, partly because they helpto reduce the long hours and demanding chores whichhave been inseparable from animal husbandry in the pastand partly because they can improve the efficiency ofproduction in substantial ways. To enlarge on these andother points it is necessary to consider the specialrequirements of each type of livestock.

2.1 PoultryChicken form the majority of UK poultry. Intensive pro-duction of broilers and eggs began in the 1950s and hasalways depended on electric power in all phases of pro-duction, from incubation of eggs, through control of feed,water and environment in growing and laying houses, toegg collection, cleaning and grading and, finally, tomanure collection [1]. Most of this power is still con-trolled by simple time switches and thermostats, althoughsince the late 1960s, electronic control of air temperaturehas been used in hatcheries and fan ventilation systemshave been used for bird housing. The latter use therm-istor sensing for control of variable-speed fan motors,rather than thermostats and on-off control. However,these have the disadvantage that at low speeds theiroperation can be affected by external winds.

Further applications of electronics on the farm had toawait the outcome of research and development in the1970s.

2.1.1 Feeding and weighing: Feed represents 75 to80% of the poultry farmer's production costs. This eco-nomic fact gave rise to two lines of research and develop-ment in the 1970s, one aimed at improved control ofrationing and the other at improved monitoring of feedutilisation by the birds.

Commercially, feed is distributed almost continuously,in compound granular form, by conveyor (Fig. 1) from

Fig. 1 Broiler house with feed conveyorsReproduced by permission of the NIAE

central hoppers of up to 10 t capacity. Distribution ratesare up to 200 g/bird/day and there may be 20 000 birds inone house [5]. Some producers now monitor this processby mounting their hoppers on electrical load cells, but,for consistent accuracies of ±3%, subhoppers of 1 tcapacity are needed. Larger enterprises are progressing tocomputer-based equipment which combines the load celloutputs with electrical outputs from water meters (eachbird drinks up to 0.2 I/day), to provide a complete historyof the birds' dietary intake.

This form of monitoring is common in otherindustries, of course, but the determination of the birds'feed/weight conversion rate presents problems of a moreunusual kind. The most obvious of these is the number ofbirds in a production unit. Clearly, they cannot all beweighed regularly. Furthermore, any attempt to catchand weigh even a small, representative sample risks thecreation of alarm or panic among them, with the conse-quent danger that they may suffer a setback in their per-formance or, worse, that some may be injured andsuffocated. For these reasons manual weighing is gener-ally carried out on a very limited scale, although moreregular information on weight gain would greatlyimprove the producer's ability to achieve target weightson specified market dates, through feed control. Regularweighing also provides a sensitive indicator of feeding,environmental and health problems.

An electronic system which overcomes this problemhas been marketed as a result of research at the NationalInstitute of Agricultural Engineering (NIAE) [6]. The keyto the system is a weighing perch which makes use of thebirds' natural behaviour to weigh them automatically,without any disturbance, to the required accuracy ofbetter than ±3%. The development of this perch raisedsome interesting design problems. In particular, it wasnecessary for the platform to be stable enough to encour-age perching but not so stable that individual birdswould be content to spend long periods on it. Overloadprotection had to be provided for the commercial loadsensor, which is a strain-gauged cantilever beam, with amaximum rating of 50 N. Then, a convenient platformheight for birds at different stages of growth had to befound experimentally. This proved to be in the range30-60 mm above floor level. The marketed design isshown in Fig. 2a. The stirrup-shaped perch is suspendedvertically, via a sprung overload protection device, fromthe cantilever beam, which is attached to one of the stan-chions in the poultry house. Some lateral sway of theperch is possible: this discourages long-stay perching.

The output of the sensor is examined by a micro-computer, which tests it for validity, rejecting any readingthat falls outside preset limits. These are set to allowthrough only those signals expected from a single per-ching bird, free from interference from others. Normallythe set limit is ±35% from a running average of acceptedweights, calculated by the computer. At any time thecomputer can display or print this average, as well as avariety of other information, including daily meanweight, weight distribution (in histogram form) and thestandard deviation of the distribution. One form ofdisplay is shown in Fig. 2b. Fig. 2c shows the resultsobtained from an automatic weighing installation overthe lifetime of a single broiler crop, with comparativeresults obtained manually. The change in growth ratebetween days 42 and 46 was due to a fault in the venti-lation system. The significant difference between manualand automatic weighings towards the market date wasascribed to the increased crowding that occurs as the


0.4-

50age of crop, days

c

Fig. 2 Automatic weighing of poultry

a Strain-gauged perchb Computer display of bird weightsc Comparison of manual and automatic weighing of a broiler crop, showing theeffect of heat stress at about 42 days# Manual weighings • automatic weighingsReproduced by permission of the NIAE

birds grow, which in this case resulted in fewer, less rep-resentative automatic weighings. However, with no morethan three perches in a 20000 bird house automatic andmanual weighings rarely differ by more than 50 g atmarket weight.

This system has also been validated with birds beingraised for layer flocks, although if these are only fed at settimes, recording should be restricted to periods whentheir body weight is relatively stable. The objective is toachieve target weights at the planned start of laying. Thesystem is equally applicable to breeder flocks, whose dietaffects the level of production of fertile eggs.

2.1.2 Environment: The health and performance of thebirds is strongly influenced by their aerial environment.Close control of air temperature through heating andventilation is therefore essential. Chicks need 27-30°C;optimum production from broilers is attained between 18and 21°C, and the optimum for layers lies between 21and 24°C. These narrow limits call for good control of airdistribution in the houses, independent of external airtemperature and wind velocity. Improved means of estab-lishing the desired ventilation pattern have been providedby an NIAE system [7] developed initially for pig houses(Fig. 3). This controls ventilation automatically byadjusting the height of an air inlet baffle which runsalong the apex of the roof, while controlling an array ofexhaust fans in the side-walls of the house. The fans areswitched on (at full speed), in sequence, as the need forventilation increases and the baffle opens progressively(step-wise) to maintain a constant 5 m/s air stream alongthe underside of the roof (which must not impede theflow). The air stream generates a stable air circulationunder a wide range of external conditions, as shown inFig. 3a. The input to the control system (Fig. 3b) is theaveraged reading of thermistor elements distributedwithin the building. Each fan is fitted with a backdraughtshutter which prevents wind interference when it isswitched off. However, as in all environmental controlsystems for closed animal houses, an unchecked rise inair temperature must lead to the emergency opening ofdoors and baffles. A separate thermostat system is ofteninstalled for this purpose.

2.1.3 Quality grading of eggs: Research at the ScottishInstitute of Agricultural Engineering (SIAE) produced asemi-automatic electronic grader for potatoes (seeSection 3.4.3) which has been adapted to egg grading.Eggs moving along a conveyor are illuminated frombelow and an operator views the light transmittedthrough them, to detect internal blood spots and otherdefects. Defective eggs are tapped with a hand-helddevice which then transmits their position to a controlunit. The latter diverts them into the reject channel.

2.7.4 Future developments: Electronic control can beusefully extended in several areas of the poultry industry,given the necessary engineering research and develop-ment. For example, precise environmental control isessential to the efficiency of the hatchery [8]. Incomingeggs pass through several incubating stages, each with itsown recommended temperature and relative humidity(RH), covering the ranges 13-5O°C and 50-75% RH,respectively. Hatchability can vary by 10%/°C change inair temperature at any stage. In addition, ventilationmust be regulated to keep oxygen levels above 15% andCO2 below 0.5% in the interests of the developingembryo. More information is needed on the environmentachieved in practice but hatcheries would almost certain-


ly benefit from the availability of low-cost O2 and CO2sensors, operating in the above ranges. These couldbecome elements in the fully integrated computer controlof hatchery operations. It would be advantageous tomonitor the RH and CO2 in broiler houses inexpensively,too, because abnormal levels of each, if prolonged,

(iii)

fans switchedin groups

electric motorfor vents

control panel

i

thermistors

ventSWI

positiontches

16

12•D

•K 8

8 12 16ventilation rate, rrf/s

c

20

Fig. 3 Automatic control of fan ventilation in livestock buildingsa Pig house (schematic), showing (i) adjustable baffle, (ii) shutter, which is closedwhen fan is not running, (iii) dunging channel and (iv) feeding passageb Control system (schematic). The vent position switches control the setting of thebaffle in relation to the number of fans running.c Ventilation requirements of a pig house over a range of ambient temperatures,calculated from the thermal insulation of the building and the number and size ofthe pigs. A 6-stage, stepped-fan system operates at equal temperature intervals, asshownReproduced by permission of NIAE

present a health risk to the birds. RH should lie between50 and 75%, while CO2 should remain near normalambient levels (0.03% by volume). The problem for elec-tronic RH sensors in this environment is the presence ofammonia (itself a potential health hazard) in the birds'litter at levels that can exceed 0.5% by weight.

Research into the environment of caged birds (layers)may reveal a need for other forms of environmental mon-itoring and control, while current studies of the

470

environment to which broilers are subjected during trans-port to the processing factories could lead to the require-ment for some form of ventilation control in thesevehicles.

Of more concern in the environmental context,though, is the apparent lack of a suitable fire sensor andalarm system to meet the safety requirements in poultryhouses and other livestock buildings [9].

Automatic weight grading of chicks is another desir-able objective, offering advantages to breeders and topoultry, meat and egg producers. The NIAE weighingperch is being used experimentally for this. More gener-ally, the perch in its existing form is likely to become partof integrated feeding, weighing and environmentalcontrol systems which will monitor the birds and regulatetheir weight gains so as to meet marketing targets.

2.2 PigsIntensive production [10] also dominates the pigindustry, which has followed — and has many parallelswith — broiler production, including the importance offeed costs (50 to 80% of total costs). Piglets are reared inheated 'creeps' at a temperature of about 35°C initially;the temperature is reduced with their size and groupnumbers thereafter. Unless they are needed for breeding,they are then grown to their market weight of up to 100kg in about six months, living in social groups of aboutten animals, fed collectively. In contrast, breeding sowsreceive individual care and rations.

A special feature of the pig industry is the problemcreated by the animals' excreta, usually called slurry,which amounts to 10 Mt annually. Many pig units haveinsufficient land on which to spread their slurry as a fer-tiliser, and many are also close enough to towns, villagesand watercourses to cause pollution problems. Thereforestored slurry is treated aerobically and/or anaerobicallyto reduce its odour and biological oxygen demand,pending its transfer to suitable land. However, otherways of using it are being explored. These will be referredto later, because they may provide new applications forelectronics.

In general, electronics has not established a strongbase in UK pig production, although in the 1970s itseemed a more promising target for research than thepoultry industry. This can be attributed to persistent eco-nomic problems in the industry. Nevertheless, there isprogress to report.

2.2.1 Feeding and weighing: Unlike intensive poultry,pigs are fed wet or dry rations (up to 3 kg/day), theformer giving better weight gains, according to manyexperts. Pig producers frequently feed their stock on anysuitable material that is available cheaply, too, ratherthan relying solely on commercial compound rations. Asa result, pig feeding systems take many forms but largerunits are likely to use labour-saving pipeline distributionsystems for either liquid or dry feed. Some of the liquidfeed systems now use microcomputers to control themixing of ingredients (water, skimmed milk, grains,chopped roots etc.) in preset proportions and to controlthe subsequent circulation of the mixture to individualpens, in metered quantities. The pipeline is in the form ofa complete loop, with a metering dispenser at each pen.Fig. 4 illustrates a commercial installation of this type forup to 200 pens. The mixing tank of up to 10 t capacitycollects the preset amount of each ingredient, under com-puter control. The computer then initiates and monitors

IEE PROCEEDINGS, Vol. 134, Pt. A, No. 6, JUNE 1987

the flow of the mixture through the pipe circuit and actu-ates each dispenser the number of times needed to deliverthe programmed feed to that pen.

Programmed, individual feeding of sows is a recentdevelopment, based on the out-of-parlour feeders fordairy cows, which will be described later. Essentially,each sow is fitted with an electronic identification unitwhich enables her to gain access to her preset ration ata communal feeding station. By this means a group ofsows can be housed in one pen and share one feeder butbe fed to their own needs.

electronic weighingsystem

Fig. 4 Automatic feeding of pigs: pipeline feeding of solid/liquid mix-tures, under computer control (schematic)Reproduced by permission of Big Dutchman

Commercially, electronic weighing of farm livestockbegan with modified mechanical weighcrates for pigs.NIAE research in the early 1970s [11] showed that thedifficulty of reading a restive animal's weight by observa-tion of the vibrating pointer on a spring balance could beovercome by simple electronic means. A linear displace-ment transducer (in this case a linear variable differentialtransformer) was mounted in parallel with the springbalance on a pig weigher and its output integrated over aweighing period of two to three seconds, to provide anaverage reading, displayed as a steady value on a digitalmeter. The weighcrate's lightly damped suspension wasfound to have a resonant frequency of about 2.5 Hz, fromwhich it could be calculated that the error due to thepig's movement during the integration period wasunlikely to exceed ±2%, unless the animal becameunusually agitated. In fact, extensive farm trials showedthat a 100 kg animal would normally be weighed with anaccuracy of ±1%. This is more than adequate, bearing inmind that daily variations in a larger pig's weight canamount to about 2 kg, due to eating, drinking and excre-tion [12]. The electronic system therefore achieved aworthwhile increase in the speed and accuracy of weigh-ing, although at additional capital cost.

Further advantages followed from improvements tothe pig handling aspects of the weighing. Pneumaticallyoperated gates, driven from a farm compressor, werefitted to the entrance and exit of the weighcrate and con-trolled by a man stationed at the entrance, as shown inFig. 5. This reduced the total weighing time per pig toabout 10 s. The overall rate of weighing was about 100pigs per hour, including the time needed to return onegroup of pigs to their pen after weighing and to bring upanother.

Despite the labour-saving features of the electronic/pneumatic system, its use has been limited so far.However, several manufacturers supply weighcrates withoptional electronic units, which use averaging techniquesand are based on load cells.


2.2.2 Environment: The system of stepped fan controlwith a variable inlet, outlined in the section on poultry,was originally developed for, and proven technically andeconomically in, intensive pig production (Fig. 3) [13].One advantage of the stable air circulation pattern pro-duced by this system is that cooler regions are establishedover dunging channels near the walls of the house,whereas it is warmer in the feeding and lying areas (Fig.3a). As pigs prefer to use the cooler regions for dunging

Fig. 5 Semi-automatic pig weighingThe weighcrate (foreground) is fitted with pneumatically-controlled inlet andoutlet gates. A linear displacement transducer (arrowed) is fitted to the springbalanceReproduced by permission of the NIAE

the result is cleaner pens and cleaner pigs, to the benefitof the animals and the farmer. However, the primaryobjective is to maintain the air temperature in the lyingarea a little above the animals' lower critical temperature(LCT). This is the temperature below which they begin touse a significant amount of the energy supplied by theirfood simply to keep warm. There is also a higher criticaltemperature (HCT) above which they use energy in anattempt to keep cool, but usually this is a minor problemin the UK. As a general indication, the LCT for pigs isnormally below 20°C and the HCT is around 30°C.However, the actual values depend on several factors. Amathematical model developed for the LCT relates it topig weight, the number of pigs per pen, the energycontent of the feed ration, the air speed over the pigs andthe type of flooring or floor covering in the pen (e.g.,straw, wood, concrete [14]. This model has been adoptedfor the control algorithm in a microprocessor-baseddevelopment of the stepped fan control system [15], asshown in Fig. 6. The Figure shows that air speed is amanual input (derived from measurements with a hand-held instrument at different ventilation settings). No suit-

471

ably robust and inexpensive sensor for the required range(0.15 to 1.0 m/s) has been found to date.

2.2.3 Ultrasonic monitoring: Ultrasonic pulse-echoequipment developed for internal flaw detection in metalswas first applied to agricultural operations by engineersin the USA in the 1960s [16]. The equipment was used todetermine the depth of backfat in live pigs, as a means ofassessing carcass quality. A 25 mm ceramic transmitter/receiver probe, energised at about 2 MHz, was pressed to

automaticinput

temperature

manual inputair flow

pig weightgroup sizefeed levelfloor type

computercontroller

managementinformationenvironment

animalperformance

fan control

varies airthroughput

inlet ventcontrol

maintainsair flowpattern

Fig. 6 Lower critical temperature control of pig environment(schematic)Reproduced by permission of the NIAE

an animal's back after smearing it with bland oil toimprove acoustic coupling. The time delays from thetransmission of pulses to the receipt of echoes fromacoustic discontinuities in the animal were determinedwith an oscilloscope. Echoes arise at the interfacesbetween lean and fatty tissue because the speed of ultra-sound in live pigs is about 1.6 km/s in lean tissue (muscle)and about 1.43 km/s in fatty tissue. The figure for pigskin is about 1.5 km/s. Developments of this form ofmonitoring have taken place in the USA, Denmark andThe Netherlands. The more recent equipment scans later-ally across the back of the immobilised animal and pro-duces an image of the corresponding cross-section of itsbody, in which the fat and lean areas are delineated.Onfarm use of this technique in the UK is mainlyrestricted to specialist advisers and breed evaluatorsbecause it requires skilled and experienced operators. Thecost of the scanning types is also too high for individualproducers.

Ultrasonics also found agricultural application in the1960s for pregnancy detection in ewes [17] and this tech-nique was subsequently adopted for sows. Pulse-echomeasurements reveal the existence of the fluid-filleduterus early in gestation and details of the developingfoetuses emerge later. More commonly, though, the foe-tuses are detected through the Doppler shift induced bythe pulsations in the sow's uterine artery or by the foetalheartbeats. The frequency shift causes the monitor toemit an audible signal. Pregnancy can be detected withinfour weeks from service, thereby enabling the producer totake timely action. Pregnant sows are given rationsappropriate to their condition and nonpregnant onesreturned to service in the next fortnight or culled asunproductive stock.

2.2.4 Future developments: Research is in progress onautomatic weighing of penned pigs, which is a prerequi-site for fully automatic control of weight gain in porkersand baconers. This can be done relatively easily if the

animals eat at a communal feeding stall, because the stallcan be fitted with a weighing platform or suspensionsystem. The programmed feeders for sows, mentionedearlier, can be adapted in this way, so enabling theassociated computer to record each sow's weight as she isidentified automatically. However, this system is verycostly for use with meat animals, because a weigher isrequired for each pen. Indirect methods of weight deter-mination, such as remote sensing of pig size, may providean acceptable, economic solution in this case.

Automatic weighing would also facilitate LCT controlof the animals' environment, which could be integratedwith feeding control in a system programmed to monitorand control growth rates, starting from ration data andtargets provided by the farmer. The development of anautomatic air-speed sensor for such a system would bebeneficial, too.

Research into new ways of using pig slurry has led topioneering farm installations of two types, namelyequipment for 2- to 3-week anaerobic digestion of theslurry, to generate biogas (methane/CO2) for farm use[18], and for vermiculture — the use of slurry as a feedfor earthworms, which are grown as a protein feed forlivestock [19]. Both processes alleviate the pollutionproblem mentioned earlier and both leave a residualcompost of agricultural value. As these processes developcommercially they are likely to require special monitor-ing and control equipment. At present the requirementsare not well defined but it is established that efficientbiogas production takes place at temperatures between30 and 35°C and that the digester should be fed at acontrolled rate with slurry of about 8% dry matter.Clearly, vermiculture is a temperature-dependent process,too.

2.3 SheepUK sheep production is still largely nonintensive, there-fore existing applications of electronics in this sector arelimited. Electronic weighing of sheep is practised, particu-larly in lowland, grain-fed flocks, but there is more inter-est in ultrasonic pregnancy detection (Fig. 7) because itcan indicate not only which ewes are pregnant but also— by the 50th day of gestation — how many foetuseseach sheep is carrying. This information enables thefarmer to feed ewes according to their needs, therebyreducing the risks of mortality among lambs and ewes.The more advanced equipment uses a linear array oftransducers, which is moved over the ewe's abdomen.The echo pattern is imaged on an oscilloscope. The tissueinterfaces can be resolved to a depth of 200 mm, using3.5 MHz transducers [20].

Looking further ahead, New Zealand workers [21]have shown that ultrasonic backfat measurements can bemade without the need for local clipping of the fleece or aconsequent loss of signal due to the twitching of thesenervous animals. Adequate coupling can be obtainedwith petroleum jelly on dry fleeces and with an aqueousgel on wet ones. Signal fluctuations are countered by fastdigitisation and averaging of the raw return signals fromthe animals. A storage oscilloscope displays and updatesthe raw and processed signals together. When the oper-ator is satisfied (on the basis of experience) that a validsignal has been received, the depth measurement is takenfrom the processed trace. Using 5 MHz, 100 ns pulses theNew Zealand workers could estimate backfat depth to±0.5 mm in the 0-10 mm range. This technique couldfind application in the UK, given the move to leanerlambs.


Closer control of feeding and of the sheep'senvironment seems unlikely, unless lowland productionand winter housing of flocks develop substantially.

2.4 Beef cattleThe UK beef sector, like the sheep industry, is mainlyassociated with low capital inputs. Therefore, again,applications of electronics are limited. Automatic controlof the environment is restricted to some calf-rearingunits, using simple electrical controls. Where automaticfeeding equipment is used this is of the kind covered inthe following Section. Weight gain can be monitored withthe aid of a larger version of the semi-automatic pigweigher (Fig. 8).

Fig. 7 Ultrasonic pregnancy detection in ewes: mobile trailer withscanning equipment

Each ewe is held in a cradle, upside down, while the real-time scan is performedReproduced by permission of the Scottish Institute of Agricultural Engineering(SIAE)

Future onfarm applications may include ultrasonic fat/lean estimation. A new method, developed at the FoodResearch Institute, Bristol [22], measures the speed oftransmission of ultrasound through the animal's hind-quarters, using a separate transmitter and receiver,mounted on a calliper frame (Fig. 9). The method pro-vides information on fat content (but not its distribution)within a few per cent and is less prone to operator errorthan the pulse-echo method, which is especially difficultwith cattle. The cost of the equipment makes it more atool for specialist advisers and breed evaluators atpresent, though. The same applies to real time scanningof cows in suckler (beef raising) herds — a techniquewhich is being developed at the Hill Farming ResearchOrganisation (HFRO) in Scotland. As in the case ofsheep, it can detect pregnancy at an early stage (about 30days post-conception) when the probe is used rectally.Repeated measurements of the size of the growing foetusalso make it possible to predict calving dates to withinthree days [23].

2.5 Dairy cattleThe UK milking herds produce not only calves for futurebeef and milk production but also over 20% of all agri-cultural and horticultural output by value. This has beenachieved by breed and herd development, coupled withcapital-intensive methods of production, involving con-tinually improving utilisation of the grass crop and con-tinually advancing technology in the dairy parlour. Theaverage size of UK dairy herds is large by Europeanstandards, at over 60 cows, and milk yields are high at

Fig. 8 Semi-automatic weighing of beef cattle

The operator is controlling the pneumatically-powered gates. The animal's weightis displayed on a meter above its head (arrowed)Reproduced by permission of the NIAE

Fig. 9 Ultrasonic measurement of animal fat, in vivo, by the transmis-

sion method

Reproduced by permission of the Food Research Institute, Bristol, UK

over 5000 kg per cow per annum. Intensive dairy pro-duction is the province of highly skilled managers andstockmen (the dairy worker is the best paid among agri-cultural workers), in which the valuable, increasingly pro-ductive cow receives a considerable measure of individualattention and treatment.

Nearly half of the national dairy herd is 'recorded' bythe UK Milk Marketing Boards [24], which collect milksamples regularly and perform quality analyses on them,including measurement of percentage butter fat andsolids-not-fat, which affect the producers' returns on themilk taken by the Boards.

Hygiene is an important factor in determining milkquality, of course, and in this context one of the dairyfarmer's continuing problems is cow mastitis, which has


serious effects on output and, if unchecked, makes theaffected cows' milk unsaleable. Regular monitoring forsigns of this infection is therefore imperative, in additionto preventive action such as regular udder and teat disin-fection.

Another important factor in the profitability of a dairyenterprise is the 'calving interval' of the herd, which is theaverage period between successive calvings by individualcows. Ideally, this should be 365 days. Should it extendto 405 days, for example, the annual yield of the herdcould be reduced by over 5%. Therefore the farmer needsto determine the onset of oestrus (heat) and pregnancy ineach cow with the greatest possible reliability.

For all of the above reasons, electronic monitoring,control and data processing has found more applicationin the dairy sector than any other in UK agriculture.

2.5.1 Feeding: Since the Second World War, grass hasbeen conserved increasingly as silage during its growingperiod, to provide a high-value bulk feed for housedcattle, either alone or in combination with additives suchas barley and minerals in granular form. Proper fermen-tation of the ensiled crop requires the exclusion of air,which quickly causes deterioration of the product. There-fore the cut crop is stored in vertical 'tower' silos or inhorizontal bunkers, clamps or bales, sealed with plasticsheets [25]. The towers are more costly but they facilitatethe production of high-quality silage, especially in regionsof high rainfall. The type of silo used is a big influence onthe mechanisation of silage feeding to cattle.

Tower silos, holding 250 t of crop or more, have topor bottom unloading mechanisms (the former requiringoccasional manual intervention to lower them as the silois emptied). Silage is always difficult to handle, because it

is moist (less than 40% dry matter), fibrous and corrosive.Unloading rates are always uneven — typically 100 kg/min maximum and 40 kg/min average, corresponding toabout 0.25 and 0.1 m3/min, respectively. Nevertheless,these rates are suitable for feed conveyor systems, whichwere first adopted in the 1960s as the basis for automa-tion of a labour-intensive operation. Pioneering develop-ment in the USA [26] was followed up in the UK at oneof the Ministry of Agriculture's experimental farms (Fig.10a) [27]. There, groups of about 60 high-yielding cowswere bulk fed with controlled amounts of silage (up to40 kg/cow/day), with rolled barley and mineral pellets (upto 5 kg/cow/day, together). Each group received a rationbased on its stage in the annual lactation cycle. Thesilage and barley for each ration were unloaded from asilo and a hopper, respectively, and then dropped intodispensers supported by electrical load cells until presetloads had been reached. Then the two ingredients weredispensed onto a conveyor, and minerals were added bya vibratory feeder for a preset time. The composition ofeach ration and its routing to the appropriate group ofcows via a conveyor network was set up manually on acentral controller. Flow and motion monitors atunloading points and on conveyors ensured correctsequential control and fail-safe operation of the system.

Subsequently, the NIAE developed a continuous beltweigher for silage [28] as part of a similar farm install-ation in which mixed rations could be routed to aninhouse conveyor/feeder (the tumbler-feeder in Fig. 10b)or to a mobile feeder wagon for external distribution.Mechanically, the weigher (Fig. 10c) was designed torestrict the clinging and corrosive silage to the uppersection of the plastic belt and to cope with flow rates upto 400 kg/min, as a precautionary measure. The 600 mm

mm~^"

rolled barleyL._ J hopper

weigher W?\ number 2 reversible. \ ypropor t ion ingauger^^ elevator/conveyor

number 3 belt S ^ = - — ^ ™M*feed w a 9 o n

conveyor •<=>- tumbler•j=|-feeder

yards 1 "v//////////7/////////////A'/////////y7/y/////////y///777/7////////////////////. '/////////7/////S/////////7/7.

Fig. 10 Automatically controlled bulk feeding of cattle, by conveyora Farm, with group housing and feeding installation for dairy cows. The picture shows two of the group houses, with forage and grain silos behind themb Farm installation with continuous belt weigher (schematic)c Belt weigherReproduced by permission of the NIAE


wide belt ran at 46 m/min. Electrically, the output fromthe weigher's 10 kg capacity load cell was used directly tocontrol the proportional metering of additives to thesilage, while the integrated signal was used to stop thefeeding when a preset weight of silage had been distrib-uted. In this way the total amounts and proportions offeedstuffs could be dispensed with an acceptable accuracyof ±5%. Although this form of mixed feed distribution islittle used in the UK at present, the method of meteringand mixing the ingredients is used in some onfarm feedpreparation plant.

The distribution of feed from horizontal silos is usuallydone with mobile machines. These include feeder wagonswith a load capacity of several tonnes, which can mixingredients loaded into them sequentially and dispensethe resulting ration along feeding troughs in the cattlelots. Some of these machines are self-propelled; othersare tractor-drawn. Either type may be fitted with an elec-tronic weighing system, based on load cells whichsupport the feed container. A digital display of the con-tainer's load during filling enables the operator tomonitor the amount of each ingredient discharged intothe wagon (Fig. 11). Indicator lamps which warn of the

Fig. 11 Controlled bulk feeding of cattle by feeder wagon

Tractor-drawn wagon with load cells and progammable meter (front of wagon)Reproduced by permission of the NIAE

approach to preset weights are sometimes fitted: thesehelp the operator to avoid overshoot. Additionally,cumulative error can be reduced by an autotaring facility,which is activated when the operator switches from thepreset level of one ingredient to the next [29]. The weightindicator can also be used to monitor discharge of theload at the feeding points.

Dairy cows are also fed individual concentrate rations,both inside and outside the milking parlour, in presetdaily amounts up to 8 kg, depending on their yield andstage of lactation. Increasingly, this is done automati-cally, with the aid of an automatic identification devicecarried by each cow, thereby saving man-hours and mini-mising human error. Out-of-parlour feeding has gainedfavour at the expense of in-parlour feeding, becausemilking times are short and research has also shown thatcows make better use of concentrates fed at intervalsthrough the day.

Although out-of-parlour feeders differ in detail theynormally have several common features, exemplified byFig. 12a. A microcomputer-based controller identifies thecow when she enters a feed stall (Fig. 12b) and lowers herhead into the manger. This action results in her identifi-cation via a transponder carried on her collar (Fig. 13a

— this will be described in Section 2.5.2). The controllerthen refers to her preset ration level and unloads a presetfraction of it from the feeder's hopper into its manger,volumetrically or gravimetrically. Most controllers cansupport 10 to 20 feeding stations, at up to 100 m range,to serve herds of 250 cows or more. The sidewalls of thestall prevent one cow from acquiring another's ration.

Fig. 12 Individual feeding of dairy cows with concentrates

a Automatic out-of-parlour feeder with transponder collar (on ledge of manger)and control unitb Feeders in useReproduced by permission of Hunday Electronics

Once a cow has eaten the dispensed fraction of herration she has to wait until the next ration period beforeshe is given more. The period can be varied throughoutthe day but is rarely less than one hour. A cow that doesnot return to the feeder in any period or periods isallowed to carry over at least part of the deficit into suc-ceeding periods. However, if she becomes an 'alarm' cow,that is, her appetite is seriously below standard, the con-troller lists her as such. This and other programmedinformation, such as the feed consumed per cow and bythe herd overall, can be displayed, printed or transmittedto the farm's central computer, where there is one [30].


energising andreceiving equipment

transponder

identitynumbertoprocessor

oscillator58kHz I

energisationamplifier r — -

"Uenergisingcoil

serialcommunicationsdecoder

pulse

receptioncoil

rectifierandsmoothing

DC supply

28Hz,

divider

906Hz. - 2nd word^ 8bits*parity

1st word8 bits• parity

supplyswitchto RFstages

24 bit shiftregister

RFoscillator26.995 MHz

code

RFmodulator

modulatedRFsignal

radioreceiver

b\T

906 Hz clock JUITUUIJ1

28 Hz load pulse J8 clock

1periods

LL

ll*

II 1

clock

MM

periods

Ml I l l l l l l

—-JUlfl

rii i

serial data t—1st word—-4—2nd word—\-complete frame

start bit •V levelstop stopbit bit

1 LSB0

MSB

'0 ' level parity bit

c

Fig. 13 Automatic identification of cowsa Collar-borne transponderb Transponder system (schematic)c Code format and timingReproduced by permission of the NIAE

2.5.2 Dairy parlour operations: Automatic feeding inthe parlour, together with automatic cow weighing andthe herd management computer are all elements of dairyparlour automation and will be covered in this Section.

In 1968 a 1300 cow dairy herd in California was fittedwith electronic equipment to record each cow's milk yieldautomatically at each milking, as a means of monitoringher physiological state [31]. This stimulated research inthe UK and The Netherlands, which began in the 1970s.It was soon realised that the key to future developmentslay in the production of a low-cost automatic identifica-tion system capable of recognising at least 250 cows indi-vidually, to cope with increasing herd sizes. The firstcommercially successful systems were based on the trans-ponders developed at NIAE (Fig. \2>b) [32]. Thesedevices, which have to conform to strict radio regula-tions, contain a ferrite-cored coil, through which thetransponder is energised by an external coil, radiating1 W maximum at 58 kHz. The coding circuit uses serialregisters to generate a 16-bit sequence (two 7-bit wordswith opposite parity), which identifies the animal con-cerned. Pulsewidth modulation of a 27 MHz carrier, at906 Hz, is used to identify the animal (Fig. 13c). Afterdecoding, the received signal passes to the parlour com-puter. A 16-bit code allows for many more identities thanan individual farmer is likely to need, but it leaves roomfor the future addition of other herd information, such asa farm code.

This identification system provided the starting pointfor integrated, computer-based monitoring and control ofdairy parlour operations on the lines defined in the UKduring the 1970s (Fig. 14) [33]. In the parlour the trans-ponder energising coil is mounted under a plastic feedmanger at each milking stall. Even where the farm usesout-of-parlour feeders it is common to give each cow atoken amount of concentrates, to encourage her to enterthe stall quickly and to relax her at the start of milking.Therefore, as each batch of animals takes up position formilking, the computer quickly identifies which individ-uals are at which stalls. Then, if the farm feeds all concen-trates in the parlour the cows are automaticallydispensed their rations for that milking. Meanwhile, aparlour worker cleans each cow's udders and teats,looking for signs of damage or inflammation, beforeattaching a vacuum milking cluster to the teats and initi-ating her milking. Her milk flow is automatically mea-sured and transmitted to the computer, which uses theinformation to determine the effective end-point of thecow's milking and her yield. At the end-point the com-puter actuates pneumatically powered automatic clusterremovers (ACRs) and the cow is ready to leave theparlour. She and the others in the batch file out whenthey are all ready, returning to their quarters via passagesor 'races'. En route they may pass through an automaticudder wash and be weighed automatically (These areoptional parts of the system).

Taking some of the elements of the system in moredetail, milk flow and yield per cow can be measured inseveral ways. If graduated recording jars (one at eachstall) are used to collect the cow's milk before it is des-patched by pipeline to the central bulk tank, the rate offlow (up to 5 kg/minute) and yield (up to about 20 kg)can be obtained by weighing each jar continuously. Fig.15a shows the base of a jar, supported on a strain-gaugedbeam. Other parlours use devices in the pipeline whichcollect, release and count small, fixed weights or volumesof each cow's milk throughout her milking. The exampleshown in Fig. \5b has a metering chamber which con-tains electrical conductivity probes at two levels, to definea volume of 0.2 1 within it and to control its charge/discharge cycle between these two levels. The meter alsoincorporates a milk sampling attachment, for MMB


recording. This and all other milk meters have to meetthe standards established by the Board. Recorded yieldmust be within ±5% or 500 g of true milk weight(whichever is the greater) in 95% of the measurementsmade. The parts in contact with the milk must be readilycleaned by (and able to withstand) the strong, hot (85°C)detergent fluids circulated through the pipelines betweenmilkings. It is fortunate that the required accuracy is nothigh, because the accuracy of weighing of recording jarsis limited by the influence of connecting pipes, whileinline meters have to deal with the pulsating flow of

frothy milk caused by the vacuum milking process.However, apart from MMB requirements, measurementof individual yields helps the farmer with herd manage-ment in various ways, including production to quota.(Note: Sales of milk are mostly based on collectionsmade by bulk tankers, which now have calibrated turbinefiowmeters with printers, to record milk transfers in litresto better than ±0.35% [34])

The determination of when a cow has finished milkingis not straightforward, because milk release tails off in agenerally exponential manner, at a rate that varies from

feeder unit

feedprocessor

parlour

keyboard anddisplay

parlourcontrolprocessor

manger coil

walk through weigherf / / /

•milkjar

I strain gauge' I ' beam

identificationtransponder

yieldprocessor

identificationprocessor

I shared memory I

aerial

radioreceiver

aerial

weigheridentification

weigherelectronics

office

managementprocessor

diskstorage

Fig. 14 Computer-based management and control system for a dairy herd (schematic)Reproduced by permission of the NIAE

Fig. 15 Automatic milk yield recordinga Recorder jar with strain-gauged support (arrowed)Reproduced by permission of the NIAEb Pipeline measuring unit, with milk sampling device (associated digital displayunit not shown)Reproduced by permission of Fullwood and Bland


cow to cow, giving milking-out times between 5 and 10minutes. To avoid undermilking some cows and over-milking others, both of which can be harmful, conven-tional ACRs operate about 30 s after a simple flowmeterhas indicated that a cow's milk flow has fallen belowabout 0.2 kg/minute. However, under computer controlthe time delay can be tailored to the cow's individualrelease rate.

The cow's udder and teats can be disinfected automati-cally, as she leaves the parlour, by a set of floor-mountedjets, actuated photoelectronically when she fully inter-rupts a horizontal array of light beams. This system isnormally independent of the computer.

Automatic, walk-through weighing can be carried outsubsequently, as the emerging batch of cows files alongthe external passage or race. The animals walk in turnover a weigh platform (which must be firm, to avoidunsettling them) and are automatically identified by theparlour computer as they pass through a transponderenergising coil at the exit end of the weigher (Fig. 16a).The computer also receives the output from the weigher'selectrical load cell (in Fig. 16a, a 2 t tension link,environmentally protected by a length of plasticdrainpipe). It relates loads to identities and rejects anyreading that shows an abnormal departure from therunning average of a cow's weight, started from a manualweighing. With cows of common size, weighing from450 to 750 kg, the maximum allowed departure is about± 30 kg, which accommodates normal daily weight varia-tions, largely due to changes in gut fill [35]. This limit

load celloutput

0 10s

Fig. 16 Automatic weighing of dairy cowsa Walk-through weigher. The cow is automatically identified as she passesthrough the wood-framed transponder energising coil at the weigher's exit.Reproduced by permission of Fullwood and Blandc Chart records of load cell outputs before and after processing (i) cow rubbingher neck on weigher (weight recorded = 612 kg) (ii) normal form of signal: 604 kgrecorded (cow's actual weight) (iii) cow chased through weigher: 585 kg recordedReproduced by permission of the NIAE

virtually eliminates readings which do not arise from thesteady passage of an animal over the platform (see fig.16b). The computer is thus able to monitor the weighttrend of most, if not all, of the herd. By this means thefarmer obtains valuable information on the animals' con-dition, usually on a weekly basis, without the expenditureof time and labour required by traditional static weigh-ings, whether electrical load cells are used or not.

All of the elements in Fig. 14 are available com-mercially, although few farmers have the complete systemat present. The only element not described so far is theparlour keyboard and display equipment through whichan operator can send information on individual cows(e.g., detected mastitis or oestrus) to the computer andreceive from it data or instructions on action to be takenin individual cases. In particular, the computer reportsfailure to receive a valid identity, or no identity, at anystall. Then, if the cause is found to be a malfunctioning orlost transponder the operator can key in the cow'snumber, read from a back-up system (collar, tail markeror freeze brand).

In the office the computer also provides the records ofindividual and herd performance — including data onyields, feeding, breeding and health — that are essentialto good management of a dairy enterprise.

2.5.3 Future developments: Research in the 1980s isconcentrating on more extensive monitoring of cattle andmilk, and on changes to milking mechanisation. The eco-nomic importance of early detection of mastitis, oestrusand pregnancy makes these the prime objectives.

Work in the USA on detection of mastitis throughconsequent change in the electrical conductivity of themilk has shown that over 80% of the cows at risk can bedetected in this way [36]. Conductivity must be mea-sured separately in all four teat attachments of themilking cluster at each milking, because infection mayfirst appear in one quarter of a cow's udder. Detectiondepends on computer processing of the resulting timeseries, taken over several days, because the effect is small(e.g., from 7 to 9 mS/cm at the normal milk temperatureof 38.5°C) and irregular, even in closely controlled experi-ments. The method shows promise but a conductivitysensor for use in the milking cluster has to be developedfor farm tests.

Research workers in The Netherlands advocate similarmonitoring of milk temperature to detect its rise atoestrus, which occurs on a 18 to 21 day cycle [37]. Thisis likely to prove more difficult, because the rise is barely0.5°C and the variations due to illness or, under normalfarm conditions, local cooling of the milk, can be of thisorder. However, many cows become more restless atoestrus, which makes it possible to detect their conditionby clamping a small, electronic activity monitor to theirforelegs [37]. The monitor is essentially a simple acceler-ometer and peak displacement counter, which can beinterrogated by an external radio receiver in the dairyparlour. The change in activity at oestrus may be 200%but it is not always so well defined. Further evaluation ofthe method is needed.

With regard to pregnancy detection and its effect on aherd's calving index, the farmer needs to know within 30days of a cow's service whether she must be returned toservice, or he may miss her next heat. Ultrasonic preg-nancy detection does not qualify on cost or earliness forthis purpose. MMB laboratory tests for progesterone inmilk samples (at the 30 nmol/1 level) identify over 98% ofnonpregnant cows about 25 days after service but this is


still marginal. Clearly, there is room for a relatively inex-pensive onfarm monitor which can shorten the period ofuncertainty. It is natural to seek a CHEMFET orBIOFET for this purpose. The same applies to theprospects for online monitoring of milk in the parlour, todetermine its lactose, fat and protein content(approximately 4.7%, 4% and 3.3%, respectively) and itsbacteriological properties. However, the MMB's centrallaboratory facilities for determination of milk quality arenow highly automated and results are normally availablein a week [38], which reduces the justification for thesemeasurements on the farm.

In the transponder sphere, miniaturisation is alreadyin progress. Reprogammable commercial units, less than50 mm square and 20 g in weight are now available inear-tag form. These can be interrogated at a range of 1 mand are suitable for both cattle and pigs. Subdermalimplanting of such transponders, in the interest ofsecurity, may follow if no insuperable difficulties ariseover their location in the animals, the range of detectionand considerations of animal welfare. Implants couldmeet the requirements of the meat processors for quick,reliable checks on the sources of their supplies.

Major changes in the mechanisation of milking mayfollow from research evidence that milk yields are greaterwhen cows are milked four or five times a day. To do thismanually would be very labour consuming, therefore theconcept of milking on demand has emerged, i.e. the cowwill be free to visit an unmanned milking station when-ever she chooses. This calls for robotic equipment toclean her udder and attach the teat cups. Work began onthis development in West Germany [39], followed by theUK and The Netherlands. If the concept proves to bepracticable the milking parlour will be replaced by agroup of concentrate feeding, weighing and milking stalls.The stockman will be free to inspect and treat cows inneed of attention, aided by information from the associ-ated computer system.

3 Arable crop and grassland production

Table 2 shows that cereals dominate all other UK arablecrops in area, yield and value. Cereal production has thelongest history of applications of electronics, too. Thisarose from the introduction of the combine harvester,which displaced the reaper/binders and threshers in the1940s and 1950s. The combine harvester enables thefarmer to harvest grain quickly during favourableweather but this calls for high-capacity grain dryinginstallations for reducing the grain moisture content to asafe storage level. Consequently, the farmer needs tomeasure grain moisture quickly and inexpensively, withan accuracy of ± 1 to 2% moisture content (wet weightbasis) in the 10 to 25% moisture content range. Elec-tronic instruments were able to meet this requirement inthe 1940s. Subsequently, electronics was applied todrying, to monitoring of the combine itself, and to seeddrilling and crop spraying earlier in the season.

Specific applications of electronics to the productionof other crops have been fewer. Automatic thinning ofsugar beet and other rowcrops, such as field lettuce, madea brief commercial appearance in the 1960s but gave wayto precision drilling or transplanting and is unlikely toreturn unless poor spring germination conditions becomea settled feature of the UK climate. Planting, harvestingand grading of potatoes has given rise to some specialisedelectronic equipment, while the grass crop provides some

recent applications to harvesting and post-harvestingoperations.

However, all field operations benefit from the monitor-ing and control equipment carried by general purposemachines and implements such as tractors and cropsprayers, while many crops have common needs in post-harvest operations. Therefore the remainder of thissection will deal with the application of electronics to theagricultural sequence of operations from cultivations toharvest and beyond, rather than crop by crop.

3.1 Cultivations and crop growth

3.1.1 Tractors: All aspects of UK field crop productioninvolve self-propelled machines. The majority of thesemachines are tractors: there are over 450 000 of them inthe UK. Few tractor functions are controlled elec-tronically at present but electrohydraulic control ofdraught implements (mainly cultivators), coupled to thetractor by its standard 3-point linkage (Fig. 17a), hasreplaced purely hydraulic control on some largemachines of 130 kW power and more. The lift angle ofthe linkage arms and the draught load on them (up to150 kN) are sensed by electrical transducers, coupled tothe electronic control units which actuate an electrohy-draulic servo-valve. Apart from reducing hysteresis andincreasing the speed of response to changing field condi-tions, the electrohydraulic system eliminates bulkymechanical couplings between the driver's cab controlsand the hydraulic lift mechanism, thereby removing con-straints on cab layout and enhancing its noise immunity.When cultivating, the driver can select 100% draughtcontrol (depth of cultivation varying), 100% depthcontrol (draught varying) or a combination of the twocontrol modes, as required by soil conditions. The sensi-tivity of the system can be varied, too [40]. The systemshown in Fig. 17a uses LVDT transducers with a 10 mmrange but recently these have been challenged by low-cost torque-sensing pivot pins and by strain-sensinginserts in the linkage arms. However, at present the costof the complete system restricts its use to largermachines.

Larger tractors are now being fitted withmicroprocessor-based monitoring equipment, too, as anaid to the driver. LED bar graph arrays display the levelof engine fuel, coolant temperature, oil pressure andbattery voltage, while LCD panels display engine speed,ground speed and the rotational speed of the machine'sPTO (mechanical power take-off for mounted or trailedimplements), when this is engaged.

Ground speed is important in many field operations.Unfortunately, it is also the most difficult to determine,because the tractor's driven wheels slip in the field, andthe speed of rotation of any nondriven wheel is notalways a sufficiently accurate basis for the determination.Nevertheless, undriven front wheels on many mediumsize tractors are fitted with magnets and their rotation isdetected by a simple magnetic field sensor, which ismounted on a wheel support (Fig. 17b). The pulse outputfrom the sensor is displayed on a ratemeter, calibrated forfield speeds (0-25 km/h, generally). The accuracy attaineddepends on speed, effective wheel diameter, wheel yawand tyre/ground pressure — the last varies with thedynamic weight distribution of the tractor/implementcombination. These limitations have led to the intro-duction of the much more costly Doppler speed meter,operating at about 10 or 24 GHz, for some speed-sensitive tasks. Fig. 17c shows a 24 GHz, 5 mW


transmitter/receiver of this type, downwardly and rear-wardly directed at an angle of 37 ± 2° to the horizontal,from about 1 m above ground level, and midmounted toreduce the influence of the tractor's rotational and verti-cal movements. The antenna produces a 5° 2-way 3 dB

(vii) (vi) (v)

( in)

(i) (ii) (iv) (v)

Fig. 17 Tractor instrumentationa Electrohydraulic implement control (schematic), showing the (i) hydraulic pumpand (ii) control valve; (iii) rams; (v) lift sensors; (iv), (vi) electronic control units;(vii) cablingReproduced by permission of Robert Boschb Ground speed measurement: magnetic wheel sensorReproduced by permission of RDS Farm Electronicsc Ground speed measurement: Doppler speed sensorReproduced by permission of TRW

beam width, with circular polarisation, and the output isa square wave at 35 Hz/km/h. Tested over a range ofagricultural surfaces in the USA [41] it achieved accu-racies of ±3% or better above 1 km/h. However, untilequipment operating at this frequency is authorised inthe UK, farmers will be restricted to the less accurate10 GHz meters.

One of the most important applications of groundspeed measurement is to the determination of wheel slip,which is quoted as the difference between the distancetravelled by a driven wheel and the distance that it wouldhave travelled under no-slip conditions, expressed as apercentage of the latter. Wheel slip is inevitable in mostfield operations but if it exceeds about 15% it brings riskof damage to soil structure, in addition to tyre wear andwaste of time and fuel. Some of the tractor monitorsmentioned above now display slip, derived from a driven-wheel tachometer and a ground speed sensor.

The driver's welfare can be at risk from slip of anotherorder, particularly during work on hillsides. In this con-nection, the SIAE developed a 'safe descent meter' whichsounds an alarm in the cab when the slope is such thatthe tractor and implement could slide downhill out ofcontrol (Fig. 18) [42]. SIAE research had shown that the

A/D convertor

dampedpendulum

brake pedalswitch

alarm

Jmicroprocessor

function selectswitches stabilised

powersupply

XT

digital displayslope in degrees

functionlamps

tractor battery

Fig. 18 Safe descent meter for work on sloping ground (schematic)

Reproduced by permission of the SIAE

slope at which an accident of this type is likely to occurcan be predicted from the deceleration measured during alocked-wheel braking test, carried out on a near-levelsurface similar in condition to that on the hillside. In thetest the pendulum sensor, which has near-criticaldamping, has an angular displacement equal to the criti-cal slope. This can lie between 15 and 30°. The micro-processor stores this information and uses it as the alarmlevel.3.1.2 Crop establishment: Soil preparation, seed drill-ing and seedling thinning are covered here, leaving refer-ence to transplanting and irrigation to the section onhorticulture.

The application of electronics to cultivation is almostentirely confined to the electrohydraulic control systemsalready described. Applications to drilling are somewhatwider. Grain drill monitors warn the tractor driver if ablockage occurs in any row of a multirow drill. A photo-cell and LED placed at the outlet of each delivery chutemonitor the flow of seeds into the soil. Cessation offlow triggers an audible alarm and the position of theblocked row is indicated by the in-cab display. Electroniclevel sensors are also used on drills to warn the driverwhen a seed hopper is almost empty.


Tramlining' is a feature of cereal drilling on manyfarms. This operation involves the suppression of seedflow in selected drill outlets, periodically, to createundrilled bands in the growing crop. These providetracks for the tractor to follow in subsequent spray andfertiliser applications. Tramlining helps to reduce over-or underlapping applications and limits the area of soilcompaction caused by field machines to the detriment ofcrop yield. Of course, the spray and fertiliser distributorsmust cover the same width as the seed drill, or a multipleof it. Usually it is the latter, therefore the tramlines arenot needed in every field bout (traverse). This has led tothe development of tramlining aids for the driver. Theseunits count the number of bouts worked by registeringthe lifting of the drill by the driver (via the drill's hydrau-lic control) during the end-of-bout turn. The driver isalerted to the tramlining bout by an audible alarm andsolenoid valves cut off the seed supply to the preselectedrows. If necessary, the driver can advance or inhibit thecount at any time, via the control unit, otherwise thelatter will regularly operate the tramlining mechanismafter a preset number of bouts (up to five in the unitshown in Fig. 19).

(iv)

© programmeadvance

(i)

Fig. 19 Tramlining aid (schematic)

(i) solenoid-actuated mechanism for suppressing seed flow in selected rows(ii) end-of-bout switch(iii) power unit(iv) control unit with audible warning (top)(v) details of panelReproduced by permission of Hestair

Turning to rowcrops, the 6-row, high-speed, precisionpotato planter developed at SIAE provides anotherapplication for ground speed measurement based on anominally nonslipping wheel. An electronic control unitchanges the speed of the planting mechanism as groundspeed varies, to maintain the planting distance at itspreset value.

Electronics applied to thinning of sugar beet and otherrowcrop seedlings had its origins in the 1930s, when a M.Ferte patented an automatic thinner which detected theposition of the seedlings with an electrical conductivityprobe, mounted horizontally and laterally to the row, justabove the soil surface. The forward travel of the machinewas measured electrically via contacts on a ground wheel.The two inputs were used to control a solenoid-actuatedblade which eliminated all seedlings in the row for apreset distance, then left unscathed the next seedling

detected. The solenoid was energised via a thyratronvalve and the whole device was powered by a tractor-mounted generator. This concept was revived in the1960s, using solid-state logic [43], but was overtaken byprecision drilling, as already stated.

3.1.3 Crop spraying and fertiliser application: Theapplication of expensive, and often highly toxic, pesti-cides, fungicides and herbicides to field crops (includinggrass) and soils is an important aspect of UK crop pro-duction. It calls for careful timing and the minimisationof waste, especially in the form of spray drift, which canresult when droplets below 200 fim in diameter are gener-ated.

Most spray equipment uses hydraulic nozzles to gen-erate the droplets, although disks are also used. Some ofboth types also have means to charge the spray electro-statically, to aid deposition on the intended target. Thecomparative efficiencies of the various systems underfarming conditions are not easy to determine but hydrau-lic nozzles seem likely to retain their dominance and theelectronic metering aids that farmers use to limit over-and under-dosing of their crops largely relate to thesedevices. Nozzles are made in a range of standard sizes,operating at pressures between 100 and 400 kPa and gen-erating droplets from 150 to over 300 /im diameter(smaller sizes at higher pressures). They are mounted atintervals along booms up to 24 m in length, to applysprays at dosages of 100 1/ha or more, depending on theirsize, at ground speeds around 8 km/h.

The simplest aids monitor ground speed and the fluidpressure in the line from the spray tank to the nozzles,the latter providing an indirect measure of the total dis-charge rate. The driver keys the width of the boom intothe monitor, then its display can be switched to show theinstantaneous ground speed, total area worked andvolume of spray applied from the start of a monitoringperiod, or the instantaneous rate of application in 1/ha. Intheory, the driver can use the instantaneous rate of appli-cation to hold the application rate at the required valueby maintaining the appropriate speed. In practice this isnot always easy, especially in the region of turns. Thisdifficulty can be overcome by the use of spray-rate con-trollers (Fig. 20a), which automatically change the dis-charge rate to counteract changes in ground speed, aswell as providing the above work data.

A constant overall rate of application does not guar-antee a uniform spray deposit, unfortunately, for severalreasons. One of these can be the difference in output fromnozzle to nozzle, possibly due to wear and tear. This canbe checked by a hand-held instrument (Fig. 206) whichuses a rotary vane flow sensor. The sensor is held up toeach nozzle in turn and the meter indicates the flow ratewith a claimed accuracy of about ± 2% in the range 0.08to 4.5 1/min. Unsatisfactory nozzles are easily replaced.

Another, more serious, cause of nonuniform applica-tion is the boom sway and whip induced by the sprayer'smovement across uneven ground. Vertical sway has beenmore than halved by the introduction of passive boomsuspensions, using springs and dampers, but it requiresan active suspension to hold the boom approximatelyparallel with and close to the ground surface [44]. Ultra-sonic height sensors at the boom tips are used to controlthe attitude of the boom via an electrically controlledadjustable link in the suspension (Fig. 20c). The trans-ducers incorporate inexpensive range sensors developedfor photographic use, which enable the controller to limittheir height differential to about 100 mm [45].


In the spraying context the Meteorological Office'scrop disease environment monitor (CDEM) assists thefarmer with the crucial timing of spray application. TheCDEM is a small weather station which computes theduration of specified combinations of temperature,humidity, rainfall and wind speed from its collected dataand, on this basis, classifies the risk of incidence ofseveral major crop diseases. Its predictions are based on

speed sensormagnetic (radar)

( i i )

driver to determine when the machine's separation andcleaning mechanisms are losing efficiency through heavyloading. The monitor detects the grain emerging from thecombine's rear waste discharge ports, where two sensorsare mounted (Fig. 21a). The amount of grain lost at therear is normally of the order of hundreds of grains persecond but this increases steeply when the threshing andcleaning stages become less effective. The sensors are

Fig. 20 Crop spraying instrumentation and control

a Spray-rate monitoring and control unit (schematic)Reproduced by permission of RDS Farm Electronicsb Hydraulic nozzle calibration unitReproduced by permission of RDSc Active boom suspension based on ultrasonic range sensors (i) traditional passiveboom suspension system (ii) Contour active boom suspension systemContour is a trademark of J.W. ChaferReproduced by permission of J.W. Chafer

crop disease-risk models developed by scientists at theUK Ministry of Agriculture, Fisheries and Food(MAFF). Other equipment of this type also uses evapo-transpiration models to compute soil moisture deficit, asan aid to irrigation scheduling [46].

The application of liquid and granular fertilisers canbe monitored and controlled by means similar to thoseemployed for crop spraying. In fact, some of the elec-tronic units used are common to all of these operations.

3.2 Crop harvestingThe expensive combine harvester is worked to the fullduring the brief, hectic grain harvest period in latesummer and early autumn. A malfunction or breakdowncan be extremely costly to the farmer. Therefore, manycombine harvesters carry engine and shaft speed moni-tors which warn the driver of sudden overloads, due tochanging crop conditions, which call for reduced speed.Many now carry a special electronic aid for avoidance ofoverload, too. This is the loss monitor, which helps the

essentially rectangular impact detectors, which can beover 3 m wide, to cover the full width of the dischargeports. The impact of grain on them produces an electricaloutput which is transmitted to a ratemeter in the driver'scab. There, the driver must decide whether the indicatedloss rate is acceptable, bearing in mind the pressure atharvest time, or whether to slow down. The meter has athreshold adjustment, to cut out background signals, anda gain control which is used to set the level of acceptableloss at midscale on its analogue display. These meters canbe used with other combined crops, such as beans, peas,oilseed rape and even grass seed.

Fig. 21a also shows the level sensors (pressure plates)which inform the driver when the grain tank is first nearand then at full capacity (usually 4-5 t). These also acti-vate the beacon which warns the driver of the attendanttractor and trailor that the combine is ready to dischargeits load.

Discharge of the harvested grain from the combineinto the trailer provides another application of elec-


tronics — the yield meter — which gives the farmer infor-mation on the variations in yield that occur in andbetween fields and different varieties of grain, under dif-ferent circumstances. Two types of meter are available.The simpler uses a magnetic sensor to count the revol-utions of the auger shaft in the combine's grain dischargetube. Constant volume discharge of grain per revolutionis assumed. The counter is calibrated by reference to thevolume or weight of a tank load of grain, determined bycalculation or use of a weighbridge, respectively.

beacon

when a metal object passes the roll. This sensor respondsto nails and short lengths of wire as well as larger objects.When an object is detected, the pick-up drive is disen-gaged by an electric clutch in 0.05 s. The driver thenreverses this drive to expose the object, shuts off the har-vester and removes the cause of the trouble. The detectoris automatically reset when the harvester is restarted.

lower feed roll

sensor assembly

amplifier

Fig. 21 Combine harvester instrumentation

a Harvester (schematic), showing position of grain loss and tank fill sensors (i)threshing mechanism (ii) straw walkers, which shake trapped grain from the straw(iii) aspirated sieves, which remove chaff and other fine contaminants from thegrain (iv) grain tankReproduced by permission of RDS Farm Electronicsb Grain yield meter mounted on the discharge auger of a combine harvesterReproduced by permission of Griffith Elder

The accuracy claimed for this simple measurement is+ 5%. A more advanced meter, with a claimed accuracyof + 2% in the range 50-200 t/h, uses a mass flow sensor,mounted at the discharge end of the auger tube (Fig.21b). The sensor consists of a load cell, supporting asloping plate, over which the grain slides before droppinginto the trailer. The output from the cell is filtered, toremove the influence of the combine's mechanical vibra-tions, then integrated to provide a measure of the mass ofgrain discharged. The meter's microcomputer also pre-dicts the weight of the grain after drying (i.e. its marketweight), given the measured value of its field moisturecontent (frequently over 20%, wet weight basis) and thetarget moisture content after drying (usually 14%, byweight), via its keyboard. This meter can be used withother combined crops.

Forage harvesting provides a different application ofelectronics. Tramp metal picked up with the crop is haz-ardous not only to the harvester but also, if undetected,to the welfare of ruminants who may ingest it. SomeAmerican forage harvesters [47] have a nonmetallicpick-up roll (Fig. 22a) fitted with an electrically balancedcoil/magnet assembly which produces an output signal

noisefi l ter alarm

Jeed roll"solenoid

thresholddetector

object rejection flap

glass fibre hood

metaldetecting coils

crop chute

choppingcylinder

high-speed pick-upwith plastic elements

crop feed system

Fig. 22 Foreign body detection on forage harvester pick-ups

a Magnetic sensor and block diagram of circuitReproduced by permission of New Hollandb Impact and magnetic sensors on high-speed pick-upReproduced by permission of the NIAE

The NIAE has developed more general foreign bodysensing for high-speed rotary forage pick-ups (Fig. 22b).The entry of any substantial object (metal or nonmetal) issensed acoustically, through its impact on the rotorcasing, and this actuates a deflector which returns theobject (with a tell-tale heap of forage) to the ground,behind the machine. The deflector quickly resets itselfand the machine does not stop. The foreign objects canbe removed from the field later. As both the rotor and thehood of the pick-up are mainly nonmetallic a detector ofsmall metal objects can also be incorporated.

Potato harvesting introduces another contaminantseparation problem in some regions of the UK, where thestones and clods gathered by the harvester can outnum-ber the potatoes in size grades over 30 mm. In the 1960sthis led the SIAE to develop a harvester which removedthese contaminants with the aid of X-rays [49]. The ele-ments of the separation system are shown in Fig. 23. Theharvested potatoes, stones and clods are dropped onto an


hydraulically levelled belt conveyor, where rollers reducethem to a single layer. On leaving the 0.7 m wide beltthey fall freely through an array of 17 X-ray beams, gen-erated by a single 30-40 kV source with 1 mA beamcurrent. A corresponding row of sensors (scintillatorscoupled to photocells) measures the attenuation of eachbeam, which is greater with the silicon-based contami-nants. The system can distinguish between 25 mm stonesand clean 150 mm potatoes — a decision taking about0.1 s. Pneumatically powered, retractable fingers returnthe stones and clods to the ground. In practice, over 90%of them can be rejected in this way.

rotating rollers

ntation belt

X-ray collimator

X-ray sensors

catching belt

Fig. 23 X-ray separator for a potato harvester (schematic)Reproduced by permission of the SIAE

3.3 Future developments in crop productionTractor monitors are slowly developing towards inte-grated monitoring and control systems, such as that out-lined in Fig. 24, designed to accept inputs from a range oftransducers on the tractor itself and on the accompany-ing implements. This requires the implement-borne codedreceive/transmit interface units shown in the Figure,which will enable the computer to recognise the functionand range of a coupled transducer. With this informationthe computer can process the transducer's output, to gen-erate standard displays and, where appropriate, thenecessary control action. This will demand a degree ofstandardisation by manufacturers of farm electronics thatdoes not exist but the agricultural engineering industry isnow debating the issue, nationally and internationally.Clearly, the manufacturers would be well advised toadopt communications standards such as RS232C as faras possible.

Existing monitoring and control systems for thetractor's 3-point linkage, seed drills and spray booms canbe merged with the above system. Additional inputs arelikely to come from safety devices, such as the SIAE pen-dulum sensor, and from torque overload sensors on thePTO (1000 kNm range). Among a variety of potentiallyuseful inputs from implements, crops and soils, contin-uous monitoring of soil moisture (in the range 10 to 50%moisture content by weight) would be of particular valuein control of drilling or planting depth. Although severalresistance and capacitance measuring probes are avail-able for soil moisture monitoring, their readings varywith soil type and condition; therefore they are best usedas undisturbed, static sensors, to indicate changes in localconditions.

The completely driverless field machine forecast in the

implementdepthmeter

flow andpressuredata

seed dri l lmonitor/control

futurerequirements

graphicdisplay

connection to^implement

keyboard

enginespeed

PTOratio

Fig. 24 Tractor and implement monitoring system {schematic)The graphic display indicates 12% wheelslip within the acceptable band (7-15%).H coded receive/transmit circuitReproduced by permission of the NIAE

alphanumeric display

firm

12°/o slip

engine speed 2242 rev/minground speed 3.6 mile/hwheel slip 12%PTO speed 1374 rev/min

microcomputer

gearratio

microwaveDoppterspeedmeter

analoguesignals


1950s [50] and researched into the 1970s [51] is still onlya future prospect. On the other hand, electronic controlsystems may contribute to the increased welfare of thedriver, through improved environmental control in cabsand noise cancellation methods for reducing cab noiselevels at frequencies below 500 Hz, which are difficult tocounter in other ways [51].

3.4 Post-harvest operationsThis is the sphere of handling, weighing, drying, gradingand storage of crops, some of which may remain in storefor months, or even years. It introduces some widely usedelectronic equipment and some less familiar outside agri-culture, particularly for moisture measurement andcontrol.

3.4.1 Weighing and handling: Vast quantities of agri-cultural produce are weighed into and out of store, either

on the move in transporters or on their way through con-veyor systems. Normally, weighings are not to tradingstandards because this is not required and the equipmentwould be too costly. However, the development of shearweb load cells has put electronic weighing platforms withan accuracy of ±0.05% or better within the economicreach of UK farmers. In consequence, axle weighers of upto 15 t capacity are now finding application on farms,together with weighbeams for supporting pallets, hoppersand other loads up to 2 t.

Fig. 25a shows a weighing platform in use with a twin-axle trailer. Weights can be measured statically ordynamically at speeds up to 3 km/h. A linked traffic lightwarns the driver if this speed limit has been exceeded.The electronic units associated with weighers of this typedisplay and print axle and total weights. Some also printdriver, vehicle, supplier or customer code numbers,acquired by manual input or automatically.

Some grain handling installations employ a constant-head, volumetric flowmeter with rotating paddles,coupled to an electronic counter. The measured bulkdensity (kg/hi; 1 m3 = 10 hi) of the grain is enteredmanually and the meter reads in tonnes, with a claimedaccuracy of ±2% up to 60 t/h. Another inline mass flow-meter, with similar accuracy, can be obtained forthroughputs up to 200 t/h. This is based on NIAEresearch in the 1970s [53] and is akin to the combineharvester discharge meter already referred to (Fig. 25b).

3.4.2 Drying: The drying of grain and other seeds tosafe storage levels is the predominant issue in this sphere.

Grain driers are broadly of two types — low tem-perature and 'high' temperature [54]. Low-temperaturedrying takes place in bulk bins (Fig. 26a) or on barn

amplifier

strain gaugebridges

activefilter

differentialamplifier

SW1-auto-zerocircuit

linearitycompensation

voltage tofrequencyconvenor

counter

Fig. 25 Static crop weighinga Installed axle weigher: 3 m wide, 10 t capacityReproduced by permission of Griffith Elderb 100 t/h mass flowmeter for grain handling installations. Sensor (schematic) and block diagram of circuit. Sensor dimensions in mmReproduced by permission of the NIAE


floors, over a period of a weak or two, using ventilatingair at or near ambient temperature [55, 56]. The air isdistributed by a fan through ducts at the base of thegrain. High-temperature driers use air heated to between40 and 100°C, depending on the crop and its intendeduse. Grain for stock feeding can be dried at 100°C butthe temperature must be much lower for milling and seedgrain. High temperature driers are of two kinds, somedrying the seed in batches and others continuously, bothtaking an hour or two or reduce the seed's moisture to itssafe storage level, near 14% moisture content by weight(Fig. 26b). Energy economy is a design and operationalobjective in all types of drying.

Low-temperature grain drying merges with the follow-ing storage phase and most of the electronic equipmentconcerned will be dealt with later. The drying stage canbe divided into three phases, programmed to avoid over-drying of the lower zone and condensation of migratingmoisture in the upper zone (see Fig. 26a). In the firstphase fan ventilation is continuous until the grain is at18% moisture content by weight. The second phase takesthe moisture content down to 16% by ventilation whenthe RH of the external air is below 70% (i.e. the equi-librium RH corresponding to 16% moisture content byweight in cereal grain). In the third phase the fan operates

air movement

only in daytime and when the air's RH is below 60%(equivalent to 14% grain moisture content by weight).

10 15 20

moisture content,*/.

b

25

Fig. 26 Grain drying and storage

a Low temperature drying and storage installation (schematic)b Safe storage limits for grainReproduced by permission of the UK Ministry of Agriculture, Fisheries and Food

100r

50

50 100

moisture content,°U wet basis

c

Fig. 27 Crop moisture meters

a Resistance typeReproduced by permission of the NIAEb Capacitance type, microprocessor-based, with hectolitre weight attachmentReproduced by permission of Sinar Agritecc Moisture determination in grass by measurement of NIR reflectance at 1300 and1450 nmReproduced by permission of the NIAE


For automatic control of low temperature drying somemanufacturers use resistive or capacitative RH sensors.Others prefer to use wet- and dry-bulb hygrometers withresistance thermometer elements. Grain moisture is mea-sured on samples from all three drying zones. The flow ofventilating air from the crop surface must be checkedwith a hand-held anemometer occasionally, too: it shouldlie between 0.1 and 0.15 m/s.

High-temperature grain drying was the main stimulusfor the development of grain moisture meters in the1940s. The Marconi moisture meter which emerged atthat time has been widely used by UK farmers since then.This early example of farm electronics was officiallytested by the NIAE, with due caution, in 1948 '. . . todetermine whether the meter is a practical piece of appar-atus for agricultural purposes', inter alia. The meter,shown in an early form in Fig. 27a, used a hand-operatedgrinder to mill a small sample of the grain, which wasthen placed in a measuring cell containing two concen-tric, stainless steel electrodes and compressed with aspring-loaded screw clamp to 1000 lb/in2 (6.9 MPa).The electrode formed part of a manually balanced, high-resistance bridge circuit, with an electrometer valve and amoving coil microammeter. Calibration scales (8-25%moisture content by weight), derived from comparisonswith standard oven measurements [57], were providedfor common types of cereals at 70°F, together with a tem-perature correction chart. Today, this and similar metersare in general use for grain and other seeds. Nowadaysthe seeds are not always milled: this matters less if thesample is taken from a batch which has reached moistureequilibrium. However, since the 1960s the capacitancemeter for whole grain has achieved increasing domin-ance. This measures the capacitance of a cell filled withgrain, at a frequency in the MHz region, to reduce theeffect of seed conductance, which depends on the mois-ture distribution within the sample (i.e. its state ofequilibrium) [58].

Fig. 21b shows a present capacitance meter for grainand other seeds, working up to 35% moisture content byweight and accepting 100-200 g samples, depending onseed size. This uses a hopper to load the cell uniformly,so reducing variations in bulk density. The loaded cell isweighed automatically. Temperature correction at about— 0.5% moisture content by weight/°C is also appliedautomatically. The microcomputer-based circuit, whichstores calibration data for seven types of seed, automati-cally displays the moisture content of the sample, cor-rected for weight and temperature, in a few seconds. Themeter also averages the results of up to 30 measurements,thereby encouraging the desirable practice of multiplesampling. A hectolitre weight (bulk density) attachment isshown in Fig. 21b. This is a simple volume measurewhich is filled with grain and then emptied into themeter's cell, where it is weighed. The meter displays thebulk density in kg/hi.

One of the earliest commercial applications of elec-tronic control to arable farming arose from the NIAE'suse of capacitance measurement to control a continuous-flow, high-temperature grain drier [59]. A capacitancesensor, with built-in temperature compensation, wasplaced in the downward moving grain stream, at a pointof minimum bulk density variation, below the dryingzone. The sensor's output, was employed to vary thethroughput rate of the grain, in accordance with a pro-portional plus integral (P + I) control algorithm.

A more recent development is the application of thenear infra-red (NIR) reflectance ratio method to the


determination of crop moisture content on the farm. Thedevelopment of LEDs which radiate at a water-absorption wavelength, such as 1450 nm, has consider-ably reduced the cost of moisture meters of this type. Incombination with a standard 1300 nm LED, which con-veniently provides the water-insensitive reference wave-length for the ratio measurement, and a germaniumphotocell to detect the reflected infra-red radiation, themeter can be used for moisture determination over awide range (Fig. 27c) [60]. The LEDs, which are ener-gised alternately, are fitted with 200 fim diameter lenses,to increase the illumination of the crop surface, which isviewed at a range of 20-40 mm. Moisture determinationtakes less than 1 ms, which makes this microprocessor-based instrument suitable for scanning the crop andaveraging the measurements very rapidly. Its obviouslimitation is that it measures surface moisture only, so inmost cases it is important for the crop to be in moistureequilibrium.

3.4.3 Storage and grading: In the arable sector elec-tronics has applications to the storage and grading ofpotatoes and to the storage of grain. Temperaturecontrol is vital in potato and grain stores. Potatoes mustbe kept above their freezing point (near — 2°C) and below10°C for long storage life [61], whereas grain tem-perature should not exceed 15°C (Fig. 26b). Temperaturesare commonly checked with hand-held digital thermom-eters, using 1 to 2 m probes, with thermistor or metalresistance sensors, for measurements at representativedepths. Automatic temperature control is also common.Here a scanned array of probes (up to 100) is installed.The layout of a control system in a potato store is shownin Fig. 28. When any of the probes detects a temperature

pressurerelief vent

potato level

internal motor,operated louvres.

airflow

^ S s ^ external5: 1) motor

stuck temperaturesensors

duct temperaturesensor

safety thermostatto air distributing system

ambienttemperaturesensor

~7an

Fig. 28 Temperature control in a potato store (schematic)

Depth of crop = 3-5 mReproduced by permission of the Electricity Council

rise above a preset level the associated controller moni-tors the ambient temperature. If the latter is cooler thanthe desired crop temperature by a preset amount(frequently, 2°C) and above the frost setting, the fan isswitched on and the louvres positioned to ventilate thecrop with external air for a set time. Should the ambientair temperature fall outside the above limits the louvresare set to circulate the internal air. Internal circulationcan also be initiated at set intervals, regardless of sensorreadings, to even out crop conditions. The controller alsologs fan running hours and it may log daily maximum,minimum and average temperatures, because super-market buyers are now interested in recorded evidence ofthe regimes under which potatoes and other food cropshave been stored.

Potato grading in large packhouses, at up to 40 t/h,has provided another outlet for X-ray sorting of potatoesfrom stones and clods, as well as for microcomputer

487

control of size and quality grading. During grading theposition of each potato on the conveyor is fixed, so it ispossible to determine its major dimensions by opticalimaging and subsequently to channel it according to itssize. The additional quality grading carried out by oper-ators on the grading line can now be performed morerapidly with the aid of a sensor developed at the SIAEand generally known as the 'magic wand' [62]. With thisin hand the operator taps any defective tuber as it passesalong the line (Fig. 29a). The impact excites a

Fig. 29 Crop grading installationsa Quality grading of potatoes with 'magic wand'Reproduced by permission of the SIAEb Automatic tomato grading line, showing individual conveyor cups (some in theforeground have swung downwards to release their contents). A colour grader ismounted over the line, in the backgroundReproduced by permission of Van Heyningen Brothers

piezoelectronic element, which causes the wand to trans-mit the position of the potato at that instant to thecontrol circuit, by inductive coupling between the wandand a matrix of fixed coils beneath the conveyor. Thisprimes a memory circuit which causes the potato to berejected downstream. The system is also being used foregg handling, as already mentioned.

A more elaborate quality grading system for potatoesuses a TV camera, mounted above the grading conveyor,to transmit a picture of the tubers to a TV monitor,where an operator (seated) taps the screen with a pen-shaped sensor wherever a defective potato appears. Thesensor picks up the monitor's line scan at that instant, sopin-pointing its position on the conveyor. This is relayedto the controller, as before, and the potato is rejecteddownstream.

3.4.4 Future developments: The increasing use of elec-tronic axle weighers and in-barn weighers will advancethe integration of data collection on the materialspassing into and out of the farm. Control of materialshandling on larger farms should provide applications forprogrammable logic controllers of the latest type.

Mathematical modelling of high temperature graindrying is developing steadily, through computer simula-tions. This may lead to new forms of microcomputer-based control, which will allow the farmer to select aparticular mode of operation, such as maximum fuel effi-ciency or minimal damage risk to seed grain.

If suitable LEDs can be developed, the NIR method ofquality determination in grain (1600-2300 nm waveband[63]) and forage (1400-2400 nm band [64]), will becomemore suitable for onfarm monitoring of these crops.Equally, research into automatic detection of defects inpotatoes (500-1900 nm band [65]) may lead to importantimprovements in the market quality of this crop.

4 Horticulture

Although the horticulture sector produces about 10% ofUK farm output by value, from about 0.2 Mha of land(Table 2), it embraces a wide range of outdoor andindoor crops, often with individual cultural requirements.No single crop dominates this sector but it is worthnoting that 2000 ha of greenhouses produces one fifth ofhorticultural output by value, and 400 ha of mushroomhouses yield nearly half that amount. In fact, protectedcrops need to be discussed separately from field veget-ables and fruit, which will be considered first. No sectionis devoted to 'ornamentals' (flowers, bulbs and shrubs)although these contribute nearly one fifth to the value ofhorticultural output. The electronics associated with theirproduction is covered by the section on protected crops,however.

4.1 Field vegetables and fruitAbout 2.5 Mt of field vegetables account for nearly onethird of the value of horticultural produce [4]. Thesecrops are normally grown on prime soil by intensivemethods, involving irrigation when and where it isadvantageous economically. Other operations specific tothis intensive production include growing and trans-planting of seedlings in compost blocks on a large scale.

Although in decline over recent years, tree and bushfruit generate about one quarter of horticultural outputby value, from about 50 kha of land. In comparison witharable farming, fruit production is still labour-intensive,like much of horticulture, and this has stimulatedresearch on planting and training the trees and bushes inways suitable for greater mechanisation.

Whatever the prospects for electronics in outdoor hor-ticulture, though, at present, it has only made a signifi-cant contribution in the post-harvest sphere of gradingand storage. Automatic thinning of lettuce and brassicasseems unlikely to challenge transplanting. In the 1960s, adriverless tractor which followed buried, AC-energisedleader cables [66] was used for routine spraying andgrass mowing in orchards. It also offered potential laboursaving in rowcrop production but it has not been devel-oped further commercially. Robotic picking of ripe fruitfrom orchard trees is an active area of research atpresent, internationally [67, 68], but its application in theUK will depend on changes in cultural practice, men-

488 1EE PROCEEDINGS, Vol. 134, Pt. A, No. 6, JUNE 1987

tioned above, and on the future size and structure of theUK fruit sector. Outdoor irrigation in the UK dependsvery little on electronic instrumentation, too. Resistiveand capacitative soil moisture sensors are available com-mercially but as their output depends on soil type andcondition, as previously noted, they need careful manage-ment under field conditions.

On the other hand, large fruit and vegetable pack-houses now employ high speed, computer-based gradingand packing lines — particularly for the supermarkets(their major customers), who demand uniformity andhigh quality. Packaging to EEC average weight stan-dards is normal. Automatic weighing on the grading lineis based on conveyors consisting of rows of hinged cups,into which individual items of produce are fed singly (Fig.29b). Each item is weighed when it reaches a load cell,which briefly bears the weight of the cup and its contents.This information is memorised by the computer and theitem is discharged at the appropriate outlet, where it ispacked with others of its grade. Quality grading is stilloften by eye and hand but in the case of tomatoes andsome tree fruit high-speed colour grading is well estab-lished. Typically, a photoelectric unit views tomatoesfrom two sides and classifies them into two or threecolour grades, covering colour ranges that can be modi-feid to meet changing market requirements. The com-bined colour/weight grader in Fig. 29b can process120000 tomatoes/h.

Storage is the other main area for electronic instru-mentation and control in this sector, as mentionedearlier. Short- and long-term storage may be used forfruit and vegetables. In the short term, the product is heldfor possibly two or three days, in a ventilated enclosureat near 0°C and 95% RH or higher, until it can be trans-ported to the retailer or food processor. This form ofstorage is now increasingly concerned with the 'cool' (or'cold') chain which starts with hydro- or vacuum-coolingimmediately after harvesting and leads on throughstorage and refrigerated transport to the supermarket.This low-temperature chain minimises all forms of dete-rioration of the produce, including water loss. However,up to now, conventional refrigeration instruments andcontrols have been used in these operations: electronicsapplications have been reserved for the long-term storageof onions, root crops and fruit. The requirement is toachieve temperatures and humidities appropriate to thecrop. For example, onions need to be stored at about 0°Cand 70-80% RH [69]. The operation of an onion store issimilar to that shown in Fig. 28 but onions, like grain,need a preliminary drying stage, with air at 70-75% RH,and heated, if necessary. Electronic thermometers andhygrometers are used, as in grain drying and storage.

The market requirement for maintenance of qualityover storage periods up to six months led to the intro-duction of gas-tight controlled atmosphere (CA) storesfor apples [70]. In these stores the CO2 level is high (1%or greater) and the oxygen level low (1-3%). Tem-perature is held at about 5°C and RH is maintainedaround 95%, to avoid serious evaporative loss from thecrop but without risking condensation on the fruit, whichcould lead to fungal or bacteriological damage. Also, ifthe oxygen level becomes too low ethanol can be gener-ated by the crop, which becomes worthless. Clearly, closecontrol of the environment is necessary. Fortunately,with the high-value crops concerned, CA storage canbear the cost of standard industrial gas analysers foroxygen and CO2, together with central monitoring andcontrol by a computer.

4.2 Protected cropsThe heated glasshouse represents another high-capitalinvestment. Running costs are considerable, too; themajor inputs are heating (usually by oil) and labour.Growers in this sector also face intense competition fromoverseas producers. Inevitably, they seek greater effi-ciency, products of higher and more uniform quality, andimproved timing of their production, to maximise theirmarket returns. Therefore, greenhouses (glass or plasticclad) are designed for maximum light transmission,although this makes them costly to heat. Also, whenautomatic control of heating and ventilation is used, thismust maintain close control of the aerial environment,despite changes in external air temperature, wind velocityand solar radiation. To meet the needs of tomatogrowers, in particular, who require temperature controlto ±0.5°C in the 15-25°C range, P + I environmentalcontrollers were adopted in the early 1960s. Control was(and still is) based on measurement of air temperature bysensors placed in aspirated screens, to protect them fromsolar radiation and consequent error [71]. Humidistats(hair type) were also introduced, to open the ridge venti-lators when the internal RH reached an upper level con-sidered to bring risk of plant disease (around 80%).External meteorological instruments were also needed toover-ride the normal ventilator control in conditions ofstrong wind or rain.

Also in the 1960s, 3-fold enrichment of aerial CO2 (i.e.,to about 0.1%) was first used in the UK to increasephotosynthesis at high light levels — for lettuce, particu-larly. Infra-red gas analysers found early application tocrop production in this way [72]. Then increased oilprices in the 1970s led to the development of 'thermalscreens', made of film plastic and drawn over and aroundthe crop at night, to achieve energy savings of 30% andmore. These were akin to the photoperiod covers alreadyused by flower growers to control the date of flowering ofsome plants.

In the plant root zone closed loop control of wateringhas been based on the electrical probes previously men-tioned, or on a tensiometer — a sealed, water-filled tubewith a porous ceramic tip, through which water is with-drawn as the surrounding soil dries, to create a measur-able vacuum [73]. Nutrient levels in watering systems aremonitored and controlled through electrical conductivityand pH meters of normal industrial pattern. This form ofcontrol is essential when the grower uses the nutrient filmor culture technique (NFT or NFC) [74] — a form ofhydroponics in which plant roots are placed in gentlysloping (about 1°) plastic channels, along which a shallowstream of nutrient liquid flows (Figs. 30a and b). Conduc-tivity is typically maintained in the 2-3 mS region (cellK = 1) and pH between 6.5 and 7, by the addition ofmixed nutrient solution and phosphoric or nitric acid,respectively.

Other forms of instrumentation are required in theheated greenhouse, to provide the grower with informa-tion helpful to efficient management. Apart from themeteorological data produced by the instruments alreadymentioned, heat input and water use in each houseshould be monitored, as should the efficiency of theboiler installation. The heat input provided by the com-monly used piped hot-water system can be determined bymeasuring the flow/return temperature difference in thepipe circuit, multiplying this by the circulation rate andintegrating the product electronically. The temperaturedifference is rarely as much as 10°C, however, and thewater often contains contaminants, including magnetic


Fig. 30 Greenhouse crop productiona Nutrient Film Technique. Young plants set out in plastic channelsb NFT installation (schematic)c Integrated computer model systems (schematic) (i) central monitoring and control

(ii) distributed monitoring and controlReproduced by permission of the NIAE

meteorologicalsensors

sensors actuatorsair ternp. steam valvepH vent motorsconductivity acid pump

salts pump

other greenhouses

floppy diskdata store

relay unit

A/D convertor

memory

processor

1st unit

analogueto digitalconvertor

digitaloutput toactuators

centralprocessor(CPU)

memory

VDU andkeyboard

data store

( i )

printer

minicomputerhost

up to 48sensors andactuators

2 wire

up to 30other units

tu p to 3km

(i i )

490 1EE PROCEEDINGS, Vol. 134, Pt. A, No. 6, JUNE 1987

particles, therefore orifice plate or venturi flow sensorsare preferred and resistance thermometers accurate to±0.1°C are essential. Then, given precise integration, anoverall accuracy of ±4% is possible [75].

Monitoring and control of heated greenhouses grewpiecemeal, inevitably, until the late 1970s, when computercontrol, developed in The Netherlands, began to appearin the UK. Fig. 30c shows, schematically, a central moni-toring and control system for a house with NFC,together with the distributed processing system, using acentral host computer, which is now favoured in the UK[76]. These systems can manage all of the functions men-tioned so far, as well as control of supplementary light-ing, where this is used. At the same time the groweracquires summarised data on weather conditions, aerialand root zone environments attained in the house, andthe consumption of fuel, water and CO2

4.3 Future developmentsAutomatic grading of fruit and vegetables can beexpected to develop, bringing in image analysis for detec-tion of abnormal shapes and other visible defects, or formore complex colour sorting. This will be equally applic-able to robotic harvesting, if required. CA storage willmake use of any reliable, low-cost sensors that becomeavailable for oxygen, CO2 and ethanol measurement. Inthe field, sensors for continuous monitoring of soil mois-ture on machines will be valuable for depth control indrilling and transplanting, as well as for irrigationcontrol.

In the protected crops sector, control of NFC solu-tions would be improved by the introduction of reliableion-specific sensors, to measure major constituents —potash (300 parts in 106); calcium and nitrogen (200 partsin 106); magnesium (40-50 parts in 106); soluble phos-phate (30-40 parts in 106) — and others, such as iron (10parts in 106). Greenhouse controllers may have to accom-modate inputs from heat pumps, wind generators andother energy sources, in addition to oil and coal-firedboilers. At a deeper level, research on computer controlof greenhouses is looking beyond P 4-1 algorithms tothose based on statistical time series analysis of theenvironment obtained in changing weather conditions,and to expert systems based on decisions made bygrowers.

One major crop — mushrooms — provides oppor-tunities to improve production through more accurateenvironmental control, in the way that tomato pro-duction has benefited over the past 40 years. Mushroomproduction proceeds in four stages, from initial pasteuris-ation of the compost, through incorporation of thespawn, subsequent spawn 'running' and on to a final,2-phase growing stage. At each step there are tem-perature constraints and in all but the first there are RHand CO2 limits. Ammonia levels are important at thefirst stage. However, the initial requirement — as was thecase in tomato production — is for research on engineer-ing of the environment, on a broad front.

5 Conclusion

Electronics has established its potential in UK agricul-ture and horticulture, through onfarm monitoring ofoperational variables, data processing and automaticcontrol, leading to the amelioration of working condi-tions, improvements in efficiency, and preservation offood quality. The cost and reliability of electronic

equipment, rather than considerations of accuracy, con-tinue to determine the rate at which farmers and growersadopt this technology, but some capital-intensive enter-prises already invest heavily in electronic equipment. Ameasure of standardisation in farm electronics will beneeded as the market increases and is beginning to takeshape.

The equipment in present use emerged from inter-actions between progressive farmers and growers, theagricultural engineering industry and multidisciplinaryresearch groups over the past decade and more. This timescale is largely dictated by the nature of farming itself,with its many variables and often harsh environment.The farm electronics of the 1990s will be determined bypresent, multidisciplinary research and development, inwhich electrical and electronic engineers have a crucialpart to play. This research and development is concen-trated first on the development or application of sensors— either for new forms of onfarm measurement or toperform existing measurements at lower cost. Second, theincreasing power of microcomputers is being exploited toproduce integrated monitoring and control systems, andto develop robotics and expert systems for agricultureand horticulture.

6 References

1 BAYETTO, R.A., PATERSON, H., and WAKEFORD, P.O.: 'Elec-tricity in agriculture' Proc. IEE, 1974,121, (11R), pp. 1249-1272

2 UK Government Statistical Service: 'Agricultural Statistics, UnitedKingdom, 1984' (HMSO, 1986)

3 UK Government Statistical Service: 'Output and utilisation of farmproduce in the United Kingdom, 1978 to 1984' (HMSO, 1985)

4 Central Statistical Office: 'Annual abstract of statistics number 122'(HMSO, 1986)

5 SAINSBURY, D.: 'Poultry health and management' (Collins Pro-fessional and Technical Books, London, 1980)

6 TURNER, M.J.B., GURNEY, P., CROWTHER, J.S.W., andSHARP, J.R.: 'An automatic weighing system for poultry' J. Agric.Eng. Res., 1984, 29, pp. 17-24

7 RANDALL, J.M.: 'A handbook on the design of a ventilationsystem for livestock buildings using step control and automaticvents'. Report 28, NIAE, Silsoe, Beds., UK

8 LUNDY, H.: 'The fertility and hatchability of the hen's egg'. UKEggs Marketing Board, 1969

9 RANDALL, J.M.: improvements in animal welfare', Agric. Eng.,1986, 41, pp. 84-90

10 WHITTEMORE, C.T.: 'Pig production: the scientific and practicalprinciples' (Longman, 1980)

11 SMITH, R.A., and TURNER, M.J.B.: 'Electronic aids for use infatstock weighing', J. Agric. Eng. Res., 1974, 19, pp. 299-311

12 TURNER, M.J.B., and COX, S.W.R.: 'Automatic weighing of live-stock'. Proceedings of National Conference on Agricultural elec-tronics applications, Chicago, IL, USA, Dec. 1983, Publication 9-84,ASAE, St. Joseph, MI, USA, 1984, pp. 563-570

13 'The Terrington integrated piggery. Advances in ventilation, feedingequipment and slurry treatment — a joint NIAE/ADAS project'.Report 42, NIAE, Silsoe, Beds., UK, 1983

14 BRUCE, J.M., and CLARK, J.J.: 'Models of heat production andcritical temperature for growing pigs'. Animal Prod., 1979, 28, pp.353-369.

15 BOON, C.R.: 'The control of climatic environment for finishingpigs', J. Agric. Eng. Res., 1984, 29, pp. 295-303

16 JOHNSON, E.K., HINES, R.L., ALSMEYER, R.H., CAMPBELL,E.L., PLATT, W.T., WEBB, J.C.: 'Ultrasonic pulse-echo measure-ment of livestock physical composition' Trans. Am. Soc. Agric. Eng.,1964, 7, pp. 246-249

17 LINDAHL, I.L.: 'Detection of pregnancy in sheep by means ofultrasound', Nature, 1966, 212, pp. 642-643

18 PAIN, B.F., HEPHERD, R.Q. (Eds.): 'Anaerobic digestion of farmwastes'. Technical bulletin 7, NIRD, Shinfield, 1985

19 PHILLIPS, V.R.: 'Engineering problems in the breakdown ofanimal wastes by earthworms'. Proceedings of International Con-ference on Earthworms in waste and environmental management,Cambridge, UK, July 1984 (NIAE, Silsoe, 1984)


20 'The determination of foetal numbers in pregnant ewes', AFRCNews, Winter 1983/84 (AFRC, London)

21 BEACH, A.D., TUCK, D.L., and TWIZELL, R.J.: 'Ultrasonicequipment for the measurement of backfat on unshorn live sheep'Ultrasonics, July 1983, pp. 184-187

22 MILES, C.A., FURSEY, G.A.J., and POMEROY, R.W.: 'Ultrasonicevaluation of cattle' Animal Prod., 1983,36, pp. 363-370

23 RUSSEL, A., and WRIGHT, I.: 'Scanning: an aid to managementin suckler herds' — in 'Science and quality in beef production'(AFRC, London, May 1986), pp. 22-23

24 The Federation of United Kingdom Milk Marketing Boards: 'Dairyfacts and figures' (Milk Marketing Board, Thames Ditton, 1985)

25 UK Ministry of Agriculture, Fisheries and Food: 'Silage'. Bulletin37 (HMSO, 1977)

26 PUCKETT, H.B.: 'Automatic livestock feed preparation and dis-tribution in the United States'. CIGR 2nd International Conferenceon Applications of electricity in agriculture, Silsoe, Beds., UK, April,1968 (CIGR, Paris)

27 TURNER, M.J.B.: 'Conveyor feeding of cattle'. Report 34, NIAE,Silsoe, Beds., UK, 1980

28 DAWSON, J.R., HOOPER, A.W., AMBLER, B.: 'A continuousweigher for agricultural use' J. Agric. Eng. Res., 1976, 21, pp.389-397

29 BRUCE, D.M., SMITH, R.A., MEEKS, I.R.: 'Mobile feedingequipment — a weighing unit with auto-taring and batching facili-ties', ibid., 1980, 25, pp. 155-160

30 'ADAS Farm business management: computing in the farm office. 3Dairy management'. Booklet B2453, UK Ministry of Agriculture,Fisheries and Food, Alnwick, Northumbs, UK, 1983

31 BRETH, F.E.: 'World's most modern dairy goes electronic', DairyFarmer, January 1968, pp. 22-25

32 STREET, M.J.: 'A pulse-code modulation system for automaticanimal identification' J. Agric. Eng. Res., 1979, 24, pp. 249-258

33 BURGESS, A.D.: 'Control, data recording and retrieval techniques'Occasional Note 2, The mechanisation and automation of cattleproduction, British Society of Animal Production, Reading, UK,1980, pp. 183-190

34 'Opening the door to flowmeters', Milk Prod., 1980, 27, (6), pp.16-17

35 FILBY, D.E., TURNER, M.J.B., and STREET, M.J.: 'A walk-through weigher for dairy cows', J. Agric. Eng. Res., 1979, 24, pp.67-78

36 PUCKETT, H.B., FERNANDO, R.S., SPAHR, S.L., and RODDA,E.D.: 'A system for analysis of milk production and conductivitydata in real time'. Proceedings of National Conference on Agricultu-ral electronics applications, Chicago, IL, USA, Dec. 1983, Pub-lication 9-84, ASAE, St. Joseph, MI, USA, 1984, pp. 508-516

37 ROSSING, W., IPEMA, A.H., KERKHOF, J.A., PLUYGERS,H.G., GARSEN, G.E., and MAATJE, K.: 'Microelectronics in dairyherd management'. Proceedings of National Conference on Agricul-tural electronics applications, Chicago, IL, USA, Dec. 1983, Pub-lication 9-84, ASAE, St. Joseph, MI, USA, 1984, pp. 606-613

38 HURST, M.: 'The analysis of milk in England and Wales for qualitypayment'. Proceedings of International Symposium on Farm elec-tronics and computing, RASE, Stoneleigh, Warks., UK, Oct. 1985,pp. 159-167

39 ORDOLFF, D.: 'Investigations on a system for automatic teat cupattaching'. Proceedings of National Conference on Agricultural elec-tronics applications, Chicago, IL, USA, Dec. 1983, Publication 9-84,ASAE, St. Joseph, MI, USA, 1984, pp. 501-507

40 HOBBS, J., and HESSE, H.: 'Electronic/hydraulic hitch control foragricultural tractors'. SAE Technical Paper Series, 801018, Societyof Automotive Engineers, Inc., Warrendale, PA, USA, 1980

41 RICHARDSON, N.A., LANNING, R.L., KOPP, K.A., and CAR-NEGIE, E.J.: 'Getting a true measure of groundspeed', Agric. Eng.,1984, 65, (6), pp. 14-20

42 SPENCER, H.B., OWEN, G.M., and GREENHILL, A.L.: 'Amicroprocessor-based safe descent slope meter', J. Agric. Eng. Res.,,1983, 28, pp. 269-272

43 COX, S.W.R., and McLEAN, K.A.: 'Electrochemical thinning ofsugar beet', ibid., 1969,14, pp. 332-343

44 MARCHANT, J.A., and FROST, A.R.: 'Spray boom attitudecontrol'. Proceedings of Agrimation 1 Conference, Chicago, IL,USA, Feb. 1985, ASAE, St. Joseph, MI, USA,

45 'Sprayer developments', Power Farming, 1986,65, (2), p. 13746 MATHIESON, I.: 'Monitoring soil moisture content for irrigation

scheduling', Soil & Water, 1986,14, (2), pp. 15-1647 BOHMAN, C.E., STIEFVATER, T.L.: 'Electronic metal detection

for forage harvesters' in 'Grain and forage harvesting'. Publication1-78, ASAE, St. Joseph, MI, USA, 1978, pp. 270-273

48 KLINNER, W.E.: 'Towards more efficient machines for harvestinggrass' Dairy Science Now, Sept. 1984, pp. 20-21 (AFRC, London)

49 PALMER, J., KITCHENMAN, A.W., MOORE, A.B., and OWEN,

G.M.: 'Adaptation of a commercial potato harvester to a com-mercial X-ray separator', J. Agric. Eng. Res., 1973, 18, pp. 355-367

50 'Driverless farm tractors' JIEE, Oct. 1959, 5, pp. 584-58651 HARRIES, G.O., and AMBLER, B.: 'Automatic ploughing: a

tractor guidance system using opto-electronic remote sensing tech-niques and a microprocessor-based controller', J. Agric. Eng. Res.,1981, 26, pp. 33-53

52 TALAMO, J.D.C., and PEACHEY, R.O.: 'Active noise control fortractor cabs?'. Proceedings of CIGR/IAAMRH/IUFRO Ergo-nomics Symposium, NIAE, 1985

53 HOOPER, A.W., and AMBLER, B.: 'A combine harvester dis-charge meter', J. Agric. Eng. Res., 1979, 24, pp. 1-10

54 ADAS: 'Grain storage methods. Grain drying and storage No. 1'.Booklet 2415, UK Ministry of Agriculture, Fisheries and Food,Alnwick, Northumb., 1982

55 ADAS: 'Bulk grain driers. Grain drying and storage No. 2'. Booklet2416, UK Ministry of Agriculture, Fisheries and Food, Alnwick,Northumb., 1982

56 ADAS: 'Managing bulk grain driers'. Booklet 2447, UK Ministry ofAgriculture, Fisheries and Food, Alnwick, Northumb., 1984

57 ADAS: 'Measuring grain moisture content. Grain drying andstorage No. 4'. Booklet 2418, UK Ministry of Agriculture, Fisheriesand Food, Alnwick, Northumb., 1985

58 MATTHEWS, J.: 'The design of an electrical capacitance-type mois-ture meter for agricultural use', J. Agric. Eng. Res., 1963, 8, pp.17-30

59 MATTHEWS, J.: 'Automatic moisture content control for graindriers', ibid., 1963, 8, pp. 207-220

60 BOWMAN, G.E.: 'The NIAE moisture meter'. Proceedings ofNational Conference on Agricultural electronics applications,Chicago, IL, USA, Dec. 1983, Publication 9-84, ASAE, St. Joseph,MI, USA, 1984, pp. 787-794

61 'Vegetable conditioning and storage — potatoes'. Technical Infor-mation AGR 8-1, The Electricity Council, Farm Electric Centre,Stoneleigh, Warks., UK, 1985

62 CARLOW, C.A.: 'An instructive rejection system for qualitygrading of potatoes and other produce', J. Agric. Eng. Res., 1983, 28,pp. 373-383

63 OSBORNE, B.G., BARRETT, G.M., CAUVAIN, S.P., andFEARN, T.: 'The determination of protein, fat and moisture inbread by near infrared reflectance spectroscopy', J. Sci. Food Agric,1984,35, pp. 940-945

64 NORRIS, K.H., BARNES, R.F., MOORE, J.E., and SHENK, J.S.:'Predicting forage quality by infrared reflectance spectroscopy', J.Animal Sci., 1976, 43, pp. 889-897

65 MUIR, A.Y., PORTEUS, R.L., and WASTIE, R.L.: 'Experiments inthe detection of incipient diseases in potato tubers by opticalmethods', J. Agric. Eng. Res., 1982, 27, pp. 131-138

66 FINN-KELCEY, P., and OWEN, V.M.: 'Leader cable tractor guid-ance'. Proceedings of Agricultural engineering Symposium, Silsoe,Beds., UK, Sept. 1967, 2/1345/18

67 PEJSA, J.H., and ORROCK, J.E.: 'Intelligent robot systems: poten-tial agricultural applications', Proceedings of 1st International Con-ference on Robotics and intelligent machines in agriculture, Tampa,FL, USA, Oct. 1983, Publication 4-84, ASAE, St. Joseph, MI, USA,1984, pp.104-111

68 GRAND D'ESNON, A.: 'Robotic harvesting of apples'. Pro-ceedings of 1st International Conference on Robotics and intelligentmachines in agriculture, Tampa, FL, USA, Oct. 1983, Publication4-84, ASAE, St. Joseph, MI, USA, 1984, pp. 112-113

69 'Vegetable conditioning and storage — onions'. Technical Informa-tion AGR 8-2, The Electricity Council, Farm Electric Centre, Stone-leigh, Warks., UK, 1981

70 SHARPLES, R., and JAMESON, J.: 'Computerised atmospherecontrol in apple stores', Fruit, Veg. & Science., Feb. 1984, (AFRC,London)

71 WINSPEAR, K.W.: 'Automation and control in glasshouses', AdaHon., 1965, 2, pp. 61-71

72 HAND, D.W., and BOWMAN, G.E.: 'Carbon dioxide assimilationmeasurement in a controlled environment glasshouse'. J. Agric. Eng.Res., 1969,14, pp. 92-99

73 BEBB, D.: 'Controlling water supply', The Grower, 1981, 95, pp.37-42

74 WINSOR, G.W., HURD, R.G., and PRICE, D. 'Nutrient FilmTechnique' Grower's Bulletin 5, Glasshouse Crops Research Insti-tute, Littlehampton, Sussex, UK, 1979

75 BAILEY, B.J.: 'Measuring glasshouse heat inputs', The Grower,1975,83, pp. 69-70

76 SANGER, T.R.: 'Measurement and control of temperature, humid-ity and carbon dioxide concentrations in greenhouses'. Proceedingsof International Symposium on Farm electronics and computing,RASE, Stoneleigh, Warks., UK, Oct. 1985, pp. 269-287


Agri

Documents

infrared reflectance

linear displacement

1st international

farm electric

achieve target

electrical

load cell

nutrient film