The case is considered of a triangular gusset plate fixed to structural members along two edges, and remaining entirely free along the third edge, which is unstiffened. The structural members are pivoted together at the intersection point of their clamping edges, that is, at the vertex of the triangular plate, and they are assumed to be stiff, and to remain practically straight when the gusset plate is deformed due to change in the angle between them. If the angle between the structural members is progressively increased, the plate will eventually commence to fail in tension near the centre of the free edge. Experiments are described for the determination of the stress distribution leading to this failure. If, on the other hand, the angle between the members is progressively decreased, then at first, stresses will be set up of opposite sign to those first mentioned, but distributed according to the same pattern. As further decrease in the angle takes place, however, transverse displacements will commence, and collapse will follow. Experiments are described for the determination of the collapse conditions. Experiments are also described relating to the collapse of plates which do not extend the whole distance from the free edge to the pivot-point of the members. The results of the experimental work are considered from a theoretical point of view, and approximate expressions are given connecting the conditions of failure with the change in angle between the members, and with the applied torque or moment about the pivot-point of the members. Whilst the methods used are applicable to plates of scalene triangular form, the work has been mainly limited to plates in the form of right-angled isosceles triangles, clamped along their equal sides, and this conforms to the majority of practical applications in engineering design.

A new liquid-liquid extractor called the oscillating Baffle Column is described. Its performance at various baffle oscillating speeds both under constant flow ratio and constant throughput runs using the system carbon tetrachloride - acetic acid - water has been investigated. Two simultaneous differential equations for the extract and raffinate phases respectively have been derived, The equations give the relationship between longitudinal mixing and mass convection in a single phase and interfacial mass transfer. By assuming equilibrium distribution to be linear, the change of Ka, the mass transfer coefficient, along the column can be represented by a polynomial. The two simultaneous equations can be solved either by the use of the Airy function' or by Power Series. The Practical applications of the solutions of these equations are limited as they are cumbersome and contain constants difficult to evaluate. A method has been derived for calculating the eddy diffusivity from concentration profile data. The calculation is restricted to the aqueous phase and has been made for a range of operating conditions. The values of the true Noy, and measured Noy, the numbers of transfer units to effect the separation differ by a 'correction term'. This correction term is a measure of the longitudinal mixing and valves for it have been calculated using a method developed. A "dimensionless fractional extraction" diagram is developed. It provides a means of measuring the operating performance of an extractor.

This research is a study of the heat transfer characteristics of cylinders with studs having constant circular cross-section and arranged in a staggered formation on the cylinder. The cylinders were placed normal to the air flow in a calibrated wind tunnel. The studs were soldered to the cylinder as a result of an initial experiment carried out to determine the merits of screwed and soldered studs. It is verified that, if the studs are not an integral part of the cylinder, there exists a resistance to heat flow at the cylinder-stud junction. Main heat transfer runs showed that as far as overall coefficient of heat transfer was concerned the length of the stud was of no consequence. A heat-transfer correlation for this case is suggested. The effect of the turbulent incident stream on heat transfer coefficient is discussed and the results of the tests conducted in the environments of the violent accoustic disturbance are presented. All the thermal runs were repeated using a plain cylinder. The variation of the heat transfer coefficient with the angular displacement of the stud from the forward stagnation point is discussed and a correlation connecting air velocity, angular position of the stud and the heat transfer coefficient is presented. This is followed by the heat transfer and pressure drop runs carried out on a studded and a plain cylinder assembly in which the width of the side walls was varied. A correlation to predict heat transfer coefficient from pressure drop measurements is suggested and verified using the results of this series of tests.

An investigation has been made into the effect of a right angled pipe bend on the heat transfer from a hot air stream to cold pipe walls. Pipe bends of various curvatures have been studied and the internal pipe diameter has been maintained constant. Reynolds numbers in the turbulent, transitional and Streamline regions have been considered and the variation of heat transfer coefficient with Reynolds number is illustrated. The test section consisted of a pipe bend and upstream and down stream straight pipes. The pipe wall was at constant temperature and the heat flux varied. A complete heat flux survey of each test section has been carried out and local heat transfer coefficients over the length and around the periphery of the pipe have been calculated. Average heat transfer coefficients have also been determined. The results have been correlated by relating heat transfer coefficients to established friction factors for curved pipe flow. The following equations are recommended for calculating heat transfer coefficients for systems containing pipe bends.

An investigation has been made into the effect of 90 and 180 bends, of 2 in internal diameter and varying diameter ratio, on the rate of heat transfer from a fluid flowing through the bend. The test section, immersed in water which acted as a heat sink, consisted of a pipe bend with upstream and downstream lengths of straight pipe. Hot air was blown through the test section and Reynolds numbers in the range 10,000 to 50,000 were studied. Local heat transfer coefficients were computed from a knowledge of local heat flux values along the pipe system. This enabled equations predicting heat transfer coefficients for a system containing a bend to be produced.

Part One describes experiments carried out on an electrically heated single cylinder mounted normal to the airstream in a closed jet wind tunnel with a square working section. Comparison of the results with previous work reveals serious discrepancies, (+20%, -40%), and these could not be accounted for by turbulence in the airstream. It is suggested that the dispersion of the results is due to the effects of the flow boundaries. After supporting this hypothesis with flow measurements the existing results are critically examined and a new correlation presented which includes the geometry of the set-up. An equation is given which correlates several workers results to within +/- 10%. Part Two describes experiments on finned cylinders. The first of these is a cylinder of diameter 2 in, to which are soldered concentric fins 4 in. diameter, 3/16 in. thick, 1/2 in. pitch. It is shown that provided the results are evaluated in a particular way, the theoretical equation for the conduction through the fins agrees closely. After a dimensional analysis of the problem, the results of the research are correlated with those of previous researches on nine different finned cylinders. Experiments in which the shape of the fins was changed are then described. It is shown that by removing material from the rear of the fins, their performance can be improved. Tests are reported which were carried out on a cylinder with eccentric fins. These fail to support the idea that such a cylinder would give better heat transfer than one with concentric fins. Further tests on the cylinder whose fins were modified reveal that two states of flow can exist over the fins. One of these results in an increase of 12% in the heat transfer. The effect is examined experimentally and suggestions to account for it are submitted.

Regulating units is process control are mostly control valves, these usually have either linear or exponential characteristic. The effect of these control valves in comparison with linear valves is investigated. The loop chosen for investigation consist a of a 3-stage RC network representing the process, a 3-term controller and a regulating unit which is considered either linear or or exponential. The characteristic of the exponential regulating unit being a non-linear one results in closed loop equations that are non-linaear differential equations. To solve these equations a numerical method which is a modification of that used by Tustin is used. The closed loop response to step changes in the desired value and load for both regulating units and under various control modes is calculated. The results obtained from the numerical solutions of the linear equations is checked by solutions using the Laplace Transform method. All results obtained from the solution of the linear and equations are checked experimentally by a simulator built especially for this purpose. Then the closed loop response to a step change in the desired value is calculated in the presence of a distance- velocity lag in the loop for both regulating units using the proportional control mode. Also the closed loop response considering saturation of the regulating units is cases of a large change in the desired value and when derivative control action is present is calculated. The results proved that an exponential regulating unit is inferior to a linear one in the case of desired value changes and is not markedly superior in case of load changes. The work is concluded by commenting on the numerical method used in the solution of the differential equations involved. It proved to be a powerful tool in dealing with non-linear control system in particular and with non-linear differential equations in general.

This thesis describes a theoretical and experimental study of the behaviour of the compression chords for through-bridges. The behaviour of the compression chords has been studied theoretically for the case when the chords have cross-sections with two axes of symmetry and negligible torsional and warping rigidities. The stability of the chords has been investigated for the cases when they remain straight up to the point of buckling , and for the cases when the live load is applied on to the flexible cross-girders and the chords consequently deflect before the buckling loads are reached. The same problems as those mentioned above have been studied for the more general case when the chords have cross-sections with only one axis of symmetry and when both the torsional and warping rigidities are taken into account. Experimental work has been carried out on a xylonite and on an aluminium model through-bridge in order to provide experimental verification for the theoretical analyses. Generally a very close agreement was obtained between the theoretical and experimental results. All the theoretical results nave been obtained using Fourier series, and they are given in the general form. Provided an infinite number of terms are used in the formulae they will give the exact answers for the problems considered. In order to estimate the practical values of these results, the convergencies of the Fourier series have been studied. It was found that for most practical cases the theoretical formulae will provide the required design data without too elaborate computations being involved.

Experiments were carried out on water flowing turbulently inside each of the five following steam heated pipes:- 1. Three copper pipes, their internal surfaces being artificially roughened by a special knurling process which produced a pattern of frasta of pyramid geometrically similar in form But varying in absolute size from pipe to pipe. The conventional roughness ratios (height of frustum of pyramid / radius of pipe) were 0.0309, 0.0228 and 0.0147. 2. A smooth copper pipe of the same inner diameter as the three artificially roughened pipes. 3. A smooth Brass pipe of 1.25 inch inside diameter. In case 1, the effect of roughness on the heat transfer and friction was investigated, and the empirical equations correlating the average heat transfer and overall roughness, By introducing the friction velocity, were deduced. Also, the average heat transfer data and friction data were correlated By assuming the eddy diffusivity for heat to Be equal to that for momentum. To show the direct effect of the height of roughness on the correlation Between Nusselt number, Prandtl number and Reynolds number, the conventional roughness ratio e/r was introduced as a parameter. The friction data for all tubes with and without heat addition are investigated. Also, a correlation Between roughness function and roughness characteristic was deduced and the equivalent sand-grain roughness was found. The results of the writer's experiments on rough tubes were correlated with those in previous research on air flowing inside tubes having square-thread type roughness and those on water flowing inside tubes having pyramid-type roughness. In Case (2) the average coefficients of heat transfer and the coefficients of friction were compared with those of the rough tubes. In Case (3) the average coefficients of heat transfer and the coefficients of friction were calculated in order to compare them with those obtained in Cases 1 and 2. In Cases 1 and 2, the efficiency of each tube was investigated By comparing the horse-power transmitted as heat with the horse-power expended in forcing the fluid through the same length of tube. In all cases the change in slope of the curve representing the relation Between Nu/Pr0.4 and Re was investigated. Finally, the effect of evaluating the physical properties of water at Bulk temperature, film temperature and surface temperature, on the equation correlating Nusselt number, Prandtl number and Reynolds number was investigated in all three cases.

An investigation has been made of fluidised systems consisting of glass beads fluidised with air. The object of the investigation was to study the spread of residence times of solids in such systems. The principal variables for investigation were:- (i) Aspect Ratio:- The length to diameter ratio of the fluidised bed. (ii) The rate of fluidising gas (Air Rate). (iii) The flow rate of solids through the system. (iv) Holdback - a parameter of mixing. (v) Segregation - a second parameter of mixing. Regression equations were produced from two factorial experiments and an attempt to correlate the results by a diffusional hypothesis was made. The expected trends were confirmed and the investigation, being one of the first of its type, opened the field for several further interesting lines of research.