This article, written by Assistant Technology Editor Karen Bybee, contains highlights of paper SPE 105449, "Design of Well Barriers To Combat Circulation Losses," by Bernt S. Aadnoy, SPE, Mesfin Belayneh, SPE, and Miguel Arriado, U. of Stavanger, and Roar Flateboe, BP plc, prepared for the 2007 SPE/IADC Drilling Conference, Amsterdam, 20–22 February. The full-length paper summarizes a 10-year research program in borehole fracturing and mud design. Numerous experiments were conducted with oil- and water-based drilling fluids to understand better the mechanisms that lead to circulation losses. A new mechanistic model for fracturing is presented that is different from other recent models, and the model was verified with laboratory experiments. The model defines optimal barrier filtrate loss to place particles in the loss zone. Selected laboratory experiments are presented that demonstrate that borehole fracturing resistance can be improved significantly by changing mud composition. Introduction Recognizing that borehole-stability mechanisms are not understood fully, research carried out during the past 10 years has focused on the fundamental physics and chemistry. Fig. 1 shows a fracturing cell where specially prepared hollow concrete cores are fractured. The setup also allows for mud circulation to ensure that mud particles are well distributed inside the hole. The cell is rated to 69 MPa, and the axial load, confining pressure, and borehole pressure can be varied independently. Many oil- and water-based drilling fluids were tested as well as novel ideas such as changing rock wettability or creating other chemical barriers. Cores with circular, oval, and triangular holes were tested to study effects of hole geometry. Fig. 2 shows typical results from the fracturing experiments. The commonly used Kirsch equation defines the theoretical fracture pressure in a nonpenetrating situation such as when using drilling muds. Only one of the three measured fracture pressures agrees with the theoretical model, and the other two are much larger. Several conclusions came out of this research.The theoretical Kirsch model underestimates the fracture pressure in general.There is significant variation in fracture pressure, depending on mud quality. This shows that the fracture pressure can be increased by designing a better mud. Several devices were constructed to study the mud and its filter cake. Fig. 3 shows a mud cell with six outlets to simulate fractures of various dimensions. The mud is circulated with a low-pressure pump to develop a filter cake across the slots. Then, a high-pressure pump increases the pressure until the mudcake breaks down. In this way, the stability and strength of the mudcake can be studied. Common muds and additives were used, and the observation was made that reducing the number of additives often results in a better mud. Nonpetroleum products also were studied to look for improvements.