Thesis (MEng)--University of Stellenbosch, 2000. / ENGLISH ABSTRACT: The compound sharp-crested weir, which consists of two or more notches at different elevations, is the most common type of flow gauging structure found in South Africa. The Department of Water Affairs and Forestry (DWAF) is responsible for the operation of these weirs. They are currently experiencing the following problems regarding flow measurement with compound sharp-crested weirs: 1. During free-flow conditions, also known as modular flows, there is uncertainty about the accuracy of the discharge formulas. Although reliable formulas exist for single notch weirs, it is not clear how they should be modified for compound weirs. Whilst methods have already been developed to deal with compound weirs their accuracy needs to be investigated further. 2. When the weirs become submerged during floods, the upstream water head is affected by downstream water levels and the original formulas are not applicable. It has so far been impossible to calculate discharge accurately under these so-called non-modular flow conditions. This report summarises the findings of an extensive study which addressed the issues mentioned above. A comprehensive test programme was completed during which variations of compound sharp-crested weirs were tested under both modular and non-modular flow conditions. In the development of new methods care was taken to adhere to internationally accepted standards. This should make South African practice acceptable to others. One important characteristic of compound weirs is the presence of end contractions. It was found that they have a significant effect on discharge measurement. Generally, the more a weir is contracted, the lower its coefficient of discharge becomes. It was also established that end contractions can ensure excellent aeration for compound weirs. For modular flow conditions it was possible to develop a discharge formula based on the IMFf equation, which is included in the ISO standards. This new method was found to produce the smallest errors when compared to other existing formulas (average error 0.6%). During non-modular flow conditions it was noted that end contractions also play an important role, at least during the early stages of submergence. In full-width weirs the upstream water level tends to drop first (due to de-aeration) before it rises again with increasing submergence. Weirs with end contractions do not suffer from this problem. It was further established that the effect of submergence could be described in terms of an energy loss occurring at the drowned weir. This energy loss is a function of the velocities at the so-called vena contracta of the weir and at the downstream river cross section. As the difference between these velocities increases, the energy loss increases as well. Two methods were tested to estimate the discharge under submerged conditions. The Villemonte formula works well if the energy losses at the weir are relatively high, while the Wessels method is reliable if the energy losses become smaller. A procedure was developed which indicates when each method is applicable. The test data indicated that the maximum error is up to ±10% at high submergence ratios (S> 0.80) if this procedure is followed. / AFRIKAANSE OPSOMMING: Multikeep-skerpkruinmeetstrukture bestaande uit twee of meer kepe op verskillende vlakke word meestal gebruik om vloeie in Suid-Afrikaanse riviere te meet. Die Departement van Waterwese en Bosbou (DWAF) is verantwoordelik vir die instandhouding van sulke meetstrukture. Hulle ervaar tans die volgende probleme met vloeimeting by multikeepskerpkruine: 1. Gedurende modulêre (onversuipte) toestande bestaan daar nog onsekerheid oor die akkuraatheid van die formules wat gebruik word om vloeitempo's te bereken. Alhoewel betroubare formules ontwikkel is vir enkelkepe, is dit nie duidelik hoe hulle aangepas kan word sodat hulle ook vir multikepe gebruik kan word nie. Die akkuraatheid van sommige bestaande metodes vir multikepe moet ook ondersoek word. 2. Onder versuipte toestande word die stroom-op watervlak beïnvloed deur die stroom-af watervlak en die formules vir vry-vloei toestande is nie geldig nie. Dit was tot nou toe nie moontlik om vloeitempo ' s akkuraat te bereken onder hierdie nie-modulêre toestande nie. Hierdie verslag bied 'n opsomming van die bevindinge van 'n navorsingsprojek wat bogenoemde probleme aangespreek het. 'n Toetsprogram is uitgevoer wat moontlike konfigurasies van multikepe ingesluit het. Die modelle van die meetstrukture is onder beide vry-vloei en versuipte toestande bestudeer. Gedurende die ontwikkeling van nuwe formules vir die berekening van vloeie is daarna gestreef om so veel as moontlik gebruik te maak van internasionale standaarde. Dit behoort Suid-Afrikaanse praktyke vir andere aanvaarbaar te maak. 'n Belangrike eienskap van multikepe is die voorkoms van end-kontraksies. Hulle het 'n groot invloed op vloeimeting en oor die algemeen het 'n keep met groot end-kontraksies 'n laer vloei koëffisient as 'n vol-wydte keep. End-kontraksies is ook baie effektief om skerpkruine te belug. Vir vry-vloei toestande is gevind dat 'n metode gebaseer op die IMFT vergelyking (ingesluit in ISO standaarde) baie goeie resultate lewer. In vergelyking met ander bestaande metodes gee dié metode die kleinste foute met 'n gemiddelde fout van 0.6%. Onder versuipte toestande speel end-kontraksies ook 'n belangrike rol, ten minste by beperkte versuiping. As vol-wydte kepe beskou word, kan gesien word dat die stroom-op watervlak eers daal voordat dit weer styg (weens ontlugting). Kepe met end-kontraksies ly nie aan hierdie probleem nie. Dit is verder bevestig dat versuipte toestande beskryf kan word in terme van 'n energie verlies wat by die versuipte meetstruktuur voorkom. Hierdie energie verlies is 'n funksie van die snelhede by die vena contraeta van die keep en by die stroom-af rivier snit. As die verskil tussen die twee snelhede groot is, is die energie velies ook groot. Twee bestaande metodes vir versuipte toestande is getoets. Villemonte se vergelyking werk goed as die energie verliese groot is, terwyl Wessels se metode beter werk as die verliese klein raak. 'n Prosedure is ontwikkel wat dit moontlik maak om die regte metode te kies vir 'n spesifieke toestand. Vanaf die toetsdata is bepaal dat die maksimum fout ±10% is by hoë grade van versuiping (S> 0.80) wanneer bogenoemde prosedure gevolg word.