Travel time estimation and prediction on motorways has long been a topic of research. Prediction modeling generally assumes that the estimation is perfect. However good the modeling, errors in estimation can significantly weaken the accuracy and reliability of the prediction. Models have been proposed for estimating travel time from loop detector data. Generally, detectors are closely spaced (say, 5(M) m), and travel time can be estimated accurately. However, detectors are not always perfect, and even during normal running conditions a few detectors malfunction, with a resultant increase in the spacing between functional detectors. Under such conditions, an error in the travel time estimation is significant and generally unacceptable. This research evaluated the in-practice travel time estimation models during various traffic conditions. Existing models fail to estimate travel time accurately under large detector spacing and during shoulder congestion periods. To address this issue, an innovative hybrid model that considered loop data for travel time estimation w as proposed. The model was tested with simulation and was validated with real Bluetooth data from the Pacific Motorway in Brisbane, Queensland, Australia. Results indicate that during non-free-flow conditions and larger detector spacing, the proposed model provides significant improvement in the accuracy of travel time estimation.