Physiological control of a left ventricular assist device (LVAD) is related to the control of blood pressure. Based on the state space model of human circulatory system, the mean aortic pressure can be estimated. This work is to study the estimation error in the presence of variations of heart rate, systole vs. diastole ratio, left ventricular contractility, resistance of blood vessels, aortic compliance. These variations may occur during the working life of a long term LVAD. Our study demonstrated that the estimator can track the large variation of resistance, and the performance depends on the accuracy of the value for an artificial parameter and the peak to peak (p-p) value of LVAD flow rate. Simulation results showed that the value of this artificial parameter is varying in different scenarios. The result of this study can be utilized for control of long term LVADs, or for non-invasive evaluation of left ventricular function and human circulatory system properties.