The paraffin (lamp oil) and heavy fuel oil refinery was established in 1888. During the Second World War it was converted into a raw mineral oil refinery for military purposes. From the end of the war till the late 1970s it produced and processed a wide range of refinery products, e.g. paste lubricants, oils, asphalt, paraffin wax or petrol. In 1965, a line for used oil reclamation was built in the refinery. The reclamation technology featured a so-called acid treatment. From 1981 the used oil reclamation remained the only manufacturing program of the refinery. In 1996 the refinery plant was closed down. The lagoon site was used to deposit the waste materials until 31 July 1996.
The diversity of products presupposed a use of various chemical substances. For example the processing of paraffin wax employed chlorinated hydrocarbons, whereas the acid treatment needed concentrated sulphuric acid. Waste generated during the mineral oil refinery production was deposited, treated/neutralised or untreated, into the lagoons.
The area concerned is therefore affected by quite diverse contamination, not only in terms of chemical composition but also physical properties. That is why the mathematical modelling itself had to be managed in several separate transport models that took account of the particular type of contamination.
The first step, common to all the models, was the compilation and calibration of the groundwater flow model in steady state and transient conditions. The model domain encompassed the right-bank floodplain of the Oder River from the catchment area of Nová Ves and Dubí down to the confluence with the Ostravice River. For certain specific simulations, this relatively large domain was limited to the so-called wider surroundings of lagoons with the western boundary of the model terminating at the edge of the over-deepened paleo-channel.
The main objective of the flow model was to assemble an “underlying” flow pattern for the subsequent transport models. These works however also included simulations of a hydraulic barrier or simulations of various pumping modes in the catchment area (CA) of Nová Ves and Dubí, taking into account the position of the groundwater divide between CA in the south and the contaminated industrial sites in the north. The latter of the aforementioned simulations were used to evaluate the degree of risk of “transmitting” the contamination to the depression cone of CA.
At this stage, the MODFLOW (saturated zone) and Hydrus (unsaturated zone for modelling the flow of water infiltrating within the remediation work into the overlaying, discontinually water bearing, backfill horizon) were applied. In the course of further steps, individual transport models were assembled, namely:
Transport model for sulphate brines
This modelling focused mainly on the so-called sulphate brines. The waste sludge from the acid treatment of the used lubricating oils was (and most likely still is) the source of highly mineralised contamination, which primarily contains sulphates but due to its low pH value is able to bind other contaminants, primarily heavy metals. High mineralisation of the solution also determines its physical-chemical properties that influence its behaviour during the transport. Due to its basic properties such as specific density higher than water and limited miscibility with fresh water the sulphate brines behave like a separate phase, usually referred to as DNAPL (Dense Non-Aqueous Phase Liquid). Given the density as mentioned above, this phase almost exclusively travels at the aquifer base and cumulates in partial depressions and channels. The groundwater flow influences the migration of brines to a limited extent only, therefore they can also expand in a direction that does not follow the groundwater flow.
For this type of contamination, the SEAWAT modelling code was used as it allows simulation of a 3D transport that reflects the density potentials. SEAWAT is a combination of MODFLOW and MT3D programs, and as such is based on the method of final differences. As its name implies, SEAWAT was primarily intended and designed for the simulation of seawater intrusion into the fresh water aquifers in the near-shore areas.
Transport models of solutes
These models, focused primarily on the area of the former Ostramo refinery, processed the following two contaminations: (1) chlorinated hydrocarbons (CHC) and (2) aromatic hydrocarbons of BTEX Group. The assembly of the transport models themselves took place in two steps. As the first step, the 1D analytical models were assembled: BIOCHLOR for chlorinated hydrocarbons, and BIOSCREEN for BTEX. The aim was to acquire the input parameters of the transport process, particularly the degradation constants of individual components. Subsequently, as the second step, the full-bodied 3D numerical models were calibrated. The input data for calibration were acquired from the monitoring. As a result the individual transport parameters were refined so that they best capture the real travel of contaminants.
For the transport model CHC, the RT3D (Reactive Transport in 3-Dimensions) model was used. It is a program code that can handle advection, dispersion and sorption, as well as modelling the 3D reactive transport of mobile and immobile substances in the saturated zone. The transport of chlorinated hydrocarbons was modelled through the sequence decomposition reaction, in the chain: PCE->TCE->1,2-cDCE->VC From the point of view of reactions, as the input parameters were defined the degradation constants and distribution coefficients – both of them for the individual components of the contamination.
The transport model BTEX was assembled in the program code MT3DMS, which allows handling the advection, dispersion and chemical reactions of the dissolved pollutants in the water-bearing systems. As regards the chemical reaction, we modelled the sorption processes and irreversible decomposition of 1st degree, as a compensation for the more complex processes of “instant” biodegradation (these were included in the calculation in the previous step for 1D transport models). The input parameters were the degradation constant and distribution coefficients – both of them for the summary parameter BTEX. Both the transport models were calibrated and validated.