Removal of initial hotspot temperature.

[Harm1995]

Since the PR that fixed the hotspot calculation error, the option to specify an initial hotspot temperature has been removed. This change was necessary because the updated hotspot calculation now consists of two components: a fast-responding heating effect from the windings and a slower cooling effect from the oil. With only a single initial hotspot temperature value, it is not possible to determine how to split this between the two components.

For a more detailed explanation of the updated formulas, see issue 32. As a result, all calculations now use the initial top-oil temperature as the starting point for the hotspot temperature.

Please let us know if this will lead to issues.

[bobduif]

I know this has only a minor impact on the results for longer time series, but I thought it might still be useful to take note of.
The loading guide (Annex I) states the following regarding the calculation of initial conditions:

Although the system may not strictly be in the steady state at the start of a calculation period, this is usally the best one can assume, and it has little effect on the result.

The initial conditions, then, are calculated by setting the time deriviaties equal to zero in each of Equations (5), (7), and (8), resulting in the following values.

From Equation (5), the initial value of $\theta_o$ is $\theta_o(0) = \left[ \frac{1 + K^2 R}{1 + R} \right]^x \times \Delta \theta_{or} + \theta_a = 63{,}9,^\circ \mathrm{C}.$

From Equation (7), the initial value of $\Delta \theta_{h1}$ is $\Delta \theta_{h1(0)} = k_{21} \times K^y \times \Delta \theta_{hr} = 53{,}2, \mathrm{K}. $

From Equation (8), the initial value of $\Delta \theta_{h2}$ is $\Delta \theta_{h2(0)} = (k_{21} - 1) \times K^y \times \Delta \theta_{hr} = 26{,}6, \mathrm{K}.$

Using this method, you can estimate the initial hot-spot (and top-oil, if needed) temperature(s) at the start of your calculations when these values are not provided. Although these estimated temperatures will most likely be higher than the actual ones, they offer a better approximation than simply assuming they are equal to the internal/ambient temperature. Additionally, with temperature limits in mind and when assessing relative aging rates, it is preferable to be on the safe side.

[Harm1995]

Thanks for the reply! This indeed sounds like an improvement to the current way of working. I will make an issue to implement this in a future version.