In this paper we discuss two different approaches to model the flexible power consumption of heat pump heated households: individual household modeling and lumped modeling. We illustrate that a benefit of individual modeling is that we can overview and optimize the complete flexibility of a heat pump portfolio. Following, we illustrate two disadvantages of individual models, namely that it requires much computational effort to optimize over a large portfolio, and second that it is difficult to accurately model the houses in certain time periods due to local disturbances. Finally, we propose a lumped model approach as an alternative to the individual models. In the lumped model, the portfolio is seen as baseline consumption superimposed with an ideal storage of limited power and energy capacity. The benefit of such a lumped model is that the computational effort of flexibility optimization is significantly reduced. Further, the individual disturbances will smooth out as the number of houses in the portfolio increases.