In order to protect chemical reactors from runaway reactions, vent sizing methods, based on data obtained from adiabatic calorimetry and two-phase flow models, were built under the guidance of the DIERS (Design Institute for Emergency System Relief). Nevertheless, because of the simplifying assumptions used (mainly conservation of initial reactive mass and homogeneous vessel venting), DIERS methods often lead to unrealistically large vent areas. This is especially the case of non tempered systems (gas-generating and hybrid systems, peroxide decomposition for example). The United Nation Manual of Test and Criteria (UN) also proposes an experimental method to determine the minimum required vent area using a 10 dm3 vessel test. The vent area is scaled up on vessel volume. Prior mass loss and two phase flow being experimentally taken into account, this method provides a more realistic vent size. This type of area to volume scale up can be conservative in the case of non tempered systems because the mass loss at small scale is more important. Nevertheless, due to the fact that the sample size is large, the UN method has the disadvantage of being time consuming, laborious and requires heavy safety precautions. This article is about the development of a new experimental vent sizing tool for non tempered system combining the advantages of DIERS method (laboratory scale) and UN method (less overconservative). This new tool consists of an extension of the adiabatic calorimeter Vent Sizing Package II (VSP2). The reactor in which chemical runaway reaction occurs is the "VSP2 blowdown test cell". In order to simulate the relief system the test cell is connected to a main vent line comprising of regulating valves which allow the simulation of "equivalent ideal orifice" area by volume ratios lying between 1024 and 5.1023 m21. A feed bleed system is also installed. The assessment of the vented mass from the test cell is done by adding at the end of the main vent line a glass column half filled with water. During the relief the chemical products are vented in the column (quench of the reaction is realised at the same time) leading to an increase in the differential pressure at the bottom of the column. The initial VSP2 device has also been modified in terms of the rate of pressure rise and fall to allow for these kinds of blowdown experiments. This new device was tested by running the decomposition of cumene hydroperoxide in 2,2,4-trimethyl-1,3-pentanediol diisobutyrate which is a non tempered system. Pressure evolution and the vented mass profile obtained are presented. The influence of several parameter (A/V ratio, the initial fill level, feed bleed opening) is also observed. However, heat losses at small scale are much more important, so working in fire simulation mode is generally preferred over the adiabatic mode. This also means that the study of tempered system appears difficult. This method can not be used with initial fill level superior to 80%. It is not adapted to violent runaway reaction (for example very concentrated peroxide solutions).