Bullshit...you cannot define the radius of a curve from the depth of the centre of a curve...you actually need a number of measurements...1) the width of the curve from each side of the straight walls it adjoins. 2) from the centre of that measurement, a number of measurements - from the centre line of that distance (across the bay to each straight wall) to intersecting points of the curve around the bay...spaced out regularly and evenly, at say 150mm distances, to establish the ‘arc’ of the curve... only then can you accurately assume the arc, and the ‘bend’ required for the radiator curve, which should match that arc...!!