Whenever cube results fail, NDT methods like ultrasonic pulse velocity, rebound hammer and core testing are used to predict in situ strength of concrete. Although common practice is to use these methods independently. Each of these methods have limitations and give inaccurate results if used individually.  

The correct practice is to use them in combination with help of multiple regression and cluster analysis.  This requires large data and use of modern statistical techniques.  

Even core testing has many pitfalls like disturbance during core extraction, method of capping and testing resulting in underestimation of in situ strengths up to 40% of actual strengths

Underestimation of in situ strengths with NDT greatly increases the cost of repairs and rehabilitation.  It also creates false alarm. 

Using advanced analysis of data from combination of test methods and correct capping technique not only give improved results but also minimize the cost of rehabilitation.