Fouling rates in refinery heat exchangers with mixed organic/inorganic deposits (frequent in practice) are estimated using a comprehensive model-based thermo-hydraulic methodology combining data-driven measurements analysis with advanced models. An industrial case study for a heat exchanger over 4-years demonstrates the method. Following an analysis of the fouling state, the dynamic analysis here estimates organic and inorganic fouling rates using constant or time-varying proportionality ratios. Base-line organics deposition rate is described by a typical correlation, inorganics deposition as a perturbation with constant or time-varying proportionality ratios.Deposition rate parameters are estimated from measured pressure drops and validated against temperatures. Results show that: the deposition rate ratio varied substantially over time, revealing acute inorganic deposition periods; accounting for inorganics explains well both thermal and hydraulic performances; the time-varying ratio provided a good fit of the data; this is a highly promising new method for predictive monitoring, detection and diagnosis of fouling. 10 ). With respect to the fouling rate, most past works focused on development and fitting of semiempirical models for chemical reaction fouling that relate dRf/dt to process conditions (e.g. 6,9,[11][12][13][14][15][16] ).Extensive reviews on the subject can be found in the literature 10,[17][18][19] . This mechanism, traditionally considered to be dominant in crude oil fouling, assumes the build-up of organic deposits over long periods of time (months, years). The thermal conductivity of such deposit has therefore been assumed to be within the range recommended for organic hydrocarbon deposits (e.g. 20,21 ), based on which, in a few modelling studies investigating the hydraulic impact of crude oil fouling (e.g. 14,22,23 ), the Rf was used to estimate the deposit thickness and its impact on the pressure drop. As discussed later in this section, ref. 24 highlighted how this approach may be not just inaccurate in predicting both fouling state and pressure drops, but also misleading (see also discussions in refs. 24-26 ).A more comprehensive method for fouling analysis based on plant data was presented by Diaz-Bejarano et al. 26 which includes two main stages: i) analysis of fouling state; ii) analysis of fouling rate. The analysis is based on applying advanced deterministic models of shell-and-tube heat exchangers undergoing fouling in combination with thermo-hydraulic measurements. The application presented is to crude oil refining, however the method is generally applicable to fouling in heat exchangers. presented an application of stage i), analysis of fouling state, to a comprehensive industrial case study involving a pre-flash, post-desalter (PDPF) heat exchanger, over a four year period, showing that extent, characteristics and transformations of fouling deposit over time could be well characterised. The combined use of temperature, flowrate and tube-side pressure drop measurements, together wi...