**1. Introduction**

The relationship between the type of bond between intersecting walls and the load-bearing capacity of test models was investigated by Castro et al. [1]. The models with no bond and with traditional masonry bond were tested. In the models without a bond, failure was caused by the loss of stability of a shorter wall component, while in the models with full bond, shearing along the whole height of the joint appeared. In addition to the joints constructed with masonry units, tests were also performed on reinforced joints. Paganoni and D'Ayala [2] investigated the effectiveness of steel anchors at the connections of intersecting walls. Similar tests were conducted by Maddaloni et al. [3,4]. However, in that case, investigations covered the effectiveness of innovative clamp anchors (rods made of carbon fibres wrapped longitudinally and spirally with a stainless steel mat). Unfortunately, there are only results from tests on connectors. Therefore, it is difficult to interpret their effectiveness because of the lack of any reference to the load-bearing capacity of joints made with a traditional masonry bond.

It is also worth mentioning tests performed by the authors in [5], in which pilot tests on masonry wall joints were presented. They were the first tests of that kind performed in Poland, and were among the few that has been performed in Europe. Within this testing programme, three types of wall joints

were compared: traditional masonry bonds, bonds with the use of steel L-shaped profiles and two-arm steel punched flat profiles. Traditional bonds exhibited almost five times higher load-bearing capacity than joints with steel L-profiles; the capacity of joints with flat profiles was almost twice higher.

The obtained test results encourage further investigations and continued work on the detailed description of joints and on the use of new methods for the construction of joints using other types of connectors, a higher number of connectors and the optimization of their shape. The performed pilot tests also demonstrated imperfections in test models and the testing method. Asymmetric failure images of two identical joints made it impossible to understand the work of a single joint. Despite the application of point forces close to the contact plane, cracks also occurred in the lower part of the web wall, which indicates bending of this part of the model and, consequently, complicates analyses. Another worrying phenomenon observed during the tests was the variation in deformations of steel connectors depending on the location of joints in relation to the loaded edge of the web wall, meaning the non-uniform work of the joints. Therefore, in further tests the authors decided to change the shape of test models and the method of load application.

The review of tests on joints described in [6] showed the lack of comprehensive studies on the behaviour of wall joints. That did not only refer to walls made of autoclaved aerated concrete (AAC) masonry units, but also made of other masonry units. A poor insight into the issue of joints and the mutual action of walls resulted in the neglect of calculations for such structures. Design standards lack guidelines for determining internal forces and stresses acting on wall intersections, and for determining conditions to verify the Ultimate Limit State (ULS) and the Serviceability Limit State (SLS). Those few tests are insufficient to describe the mechanism of joints work, much less to develop guidelines for their design and construction. Moreover, there is the need to design a connector in a new shape to satisfy the demands of the market, which is aimed at optimizing existing solutions. A new connector should meet requirements of ultimate states and simultaneously should have a simple construction, easy assembly and much higher performance reliability in the phase after reaching the greatest loading. Therefore, the overall aims of our own tests were specified and they included:


Moreover, the authors made an attempt to build simplified models representing the behaviour of reinforced and unreinforced joints, a process which was described in this paper and in [7]. Tests and analyses presented in this paper were completed with a new series of tests.
