The characteristics of the particle distribution in a sonic jet release of carbon dioxide (CO2) from a high pressure reservoir are investigated. The motivations are to quantify the hitherto unknown particle distribution immediately after the Mach shock and examine the level of agglomeration in the jet post-Mach shock. These releases are designed to be representative of a sonic release into the atmosphere and so provide data to help interpret how accidental or deliberate releases from the carbon capture and storage (CCS) chain might behave. We discuss the experimental results in comparison with numerical work, focusing on implementing the initial particle distribution and reproducing the observed agglomeration in a Reynolds-averaged Navier-Stokes numerical model with an adaptive grid, and in terms of the particle distribution function, Lagrangian particle tracking and turbulent shear agglomeration. INTRODUCTION Predicting the correct fluid phase and solid particle behaviour during the discharge process in the near-field of sonic CO2 jets is of particular importance in assessing the risks associated with carbon capture and storage schemes, given the very different hazard profiles of CO2 in the gaseous and solid states. The initial particle size formed in releases from the liquid phase at high pressure and temperature is unknown. Large-scale experiments and accompanying simulations have indicated that agglomeration is occurring along the sonic jet formed following a free release into the atmosphere. This work was commissioned by National Grid as part of the COOLTRANS research programme in order to experimentally investigate the particle size distribution and agglomeration. National Grid has initiated this programme to address knowledge gaps relating to the safe design and operation of onshore pipelines for transporting dense phase CO2 from industrial emitters in the UK to storage sites offshore. EXPERIMENTAL MEASUREMENTS The experiment was conducted in a laboratory setting in a large container with a separate vent system fitted to ensure safe handling of the CO2. A 20ml canister of liquid CO2 was pressurised to 68.9bar and allowed to equilibrate to ambient temperature for one hour. The canister was then clamped into a frame with the nozzle protruding into a custom- made Perspex box (5050500 mm), flush with the internal surface of the box. Two custom-made nozzles were used with diameters of 0.5mm and 1mm. The instrument used for measurement was a Dantec fiberflow laser Doppler anemometer (LDA), with a Dantec classic phase Doppler anemometer (PDA) module. The data were processed using a Dantec burst spectrum analyser and Dantec BSA flow software. The illumination was provided by a Spectra-Physics Stabilite 2017 multi-spectral argon-ion continuous wave laser. The LDA was initiated and the measurement volume was located in line with the centre of the jet, at a range of distances from the nozzle, measured in nozzle diameters. Data collection was commenced and 10 seconds later a 1/4 turn gas valve was opened to release the CO2 from the canister into the Perspex box. Each experiment was released into the atmosphere in the container, mimicking a discharge into a regular atmosphere. The base case employs a 0.5mm diameter nozzle, with measurements obtained at 6, 10, 20, 30, 50, 100 and 150 diameters (D) downstream. A 1mm diameter nozzle was also used, with measurements obtained at 3, 5, 10, 20, 30, 50, 100 and 150 D.