Has anyone made a serious attempt to apply option theory to real assets and capital structures, taking into account all the messy details ?
If you imply real assets as the "left" side of the balance sheet, then there are recent papers which examine simultaneously both the option to invest (growth options), issuance of debt/equity and the ability to determine endogenously the default barrier. Although, their complexity may require numerical solutions and increased mathematical complexity that might repel practioners. Another issue might be high sensitivity to initial parameters, such as bankruptcy costs, recovery rate on debt, volatility, etc.
From the family of capital structure models, the most widely known is the industrial version of Merton's model, called KMV. Practioners need to determine the Probability of Default to assess credit risk for monitoring purposes. Using a proprietary database, distance to default is mapped to expected default frequency.
For determining the optimal capital structure, someone could use, for instance, the models of Leland or Leland & Toft to find the optimal leverage that maximizes firm's value. Although, whether industry uses such models to determine capital structure, remains questionable.