Jul-2006

Refinery CO2 challenges: part III

For the prediction of CO2 emissions from a refinery, simple correlations are not always sufficient. A rigorous simulation tool that includes fractionation and reactor models can help to obtain a correct prediction of the total emissions

Joris N Mertens, Klaas Minks and R Michiel Spoor, KBC Process Technology

Article Summary

In part I of this series of articles on refinery CO2 emissions, the challenges facing the refiner regarding CO2 emissions reduction were discussed.1 One of the biggest challenges is the calculation/prediction of CO2 emissions. Although the calculation of actual emissions by measuring and monitoring fuel consumption on fired heaters and boilers is relatively straightforward, to actually predict emissions requires more effort. The total amount of CO2 emissions depends on process unit throughput and energy efficiency, but crude oil slate, process unit constraints, product qualities and refinery infrastructure play an important role (Figure 1) too. These factors may also determine the total emissions and so need to be taken into consideration when calculating CO2 emissions.

First approach for CO2 emissions prediction
The first approach for the prediction of refinery emissions involves simplified correlations between feed properties and the mode of operation for each refinery unit. Correlations of this type have been developed by KBC in the past as part of their Best Technology (BT) efficiency benchmark.2 For example, a crude distillation unit (CDU) will consume a certain amount of fuel in the heaters based on the residue volume yield. With higher volumes of residue, the total energy consumption of the unit per tonne of feed reduces. Figure 2 shows the relationship between residue yield and fuel consumption. When applying the actual unit BT efficiency to the correlations, a reasonable estimate of furnace duty and CO2 emissions can be made for that particular unit.

These calculations are extremely useful for implementing in a refinery LP planning system to forecast refinery CO2 emissions based on simple correlations. However, this approach is not sufficiently accurate for the following cases:
— When infrastructural changes need to be made (new projects)
— Modelling utility CO2 contribution and the effect of utility infrastructure on total emissions
— Modelling indirect CO2 emissions from fuel balance and unit steam/power consumption
—  For non-standard refinery units where existing correlations are not applicable or for non-standard configurations of units
— For variations in feed properties that are not part of the existing correlations but do have an effect (direct or indirect) on CO2 emissions. For example, methane/ethane yield from the crude oil affects CO2 emissions from the marginal fuel on the refinery.

Therefore, rigorous modelling of the distillation columns, conversion reactors and heater duty calculations is required in combination with site-wide utility modelling to achieve detailed and accurate CO2 emissions forecasting. For both, rigorous modelling of the refinery and utility/infrastructure system in an integrated approach is essential. For the catalytic reformer and FCC units, two of the main CO2-producing refinery units, the effect of external factors (for example, feed quality and unit/product constraints) and infrastructure on CO2 emissions is discussed. For this article, other large CO2 producers, such as the crude distillation and vacuum units, have been left out of the discussion.

Naphtha reforming
CO2 emissions generated by naphtha reformers consist of direct emissions by the reforming unit itself and indirect emissions that do not occur on the reforming unit but affect the CO2 balance of the refinery elsewhere. Direct CO2 emissions from naphtha reformers are a result of furnace firing. In addition to the main furnace, LPG recovery requires energy. Although most units have steam reboiled debutanisers, some use reboiler furnaces and others use reactor effluent heat. The main variables that affect reformer furnace duty are:
— Heat required by the reforming reactions. The heat required is largely set by feed quality and operating severity (product octane)
— Hydrogen recycle rate
— Overall and radiant furnace efficiency
— Heat integration. Higher losses of process heat to air or water increase the fuel demand.

To estimate CO2 emissions and investigate how operating variables affect CO2 emissions, two existing naphtha reformers have been modelled using KBC’s proprietary REF-SIM software. The first unit is a state-of-the-art low-pressure CCR (Figure 3); the second unit is an old high-pressure “semi-regen” unit operating with a much lower hydrogen and reformate yield, a high gas and LPG yield, and poor heat integration. Table 1 shows the main operating variables of the units. These two units have been selected because they are at the extremes of the operating range of naphtha reformers. Both units use the radiant section of the furnaces to generate process heat and the convection section to produce steam.

The radiant heat absorbed is the energy required to preheat the feed mix added to the inter-heater radiant duty. A much higher duty would be expected for the CCR unit because of the higher endotherm. However, due to the poor heat recovery in the feed/effluent exchanger of the old semi-regen unit, this potential disadvantage of the CCR design is largely offset. In addition, the higher recycle gas ratio further increases fuel demand on the higher-pressure unit.

Fuel oil generates 0.33 tonne of CO2 per GCal of fuel oil fired, while CO2 emissions from natural gas are limited to 0.23 tonne/Gcal.1,3 This results in a specific direct CO2 emission on the furnace of 145–165kg of CO2 per tonne of feed, assuming natural gas is being fired. Thus, at 100 tonnes per hour of feed (20MBD), both units will generate some 130 ktonnes of CO2 each year.

However, to get the complete picture, indirect emissions should be accounted for as well. First, the fuel fired on the reformer furnace itself will have an indirect impact on carbon emissions if the marginal fuel gas differs from the fuel fired on the unit, which will often be the case. Fuel oil is the more typical marginal fuel in Europe. Consequently, in addition to the 145/165kg of CO2 produced per tonne of feed, CO2 emissions will increase elsewhere in the refinery if the reformer fires natural gas but uses fuel oil as the marginal refinery fuel. This is due to the fact that burners on other units will have to be switched from natural gas to fuel oil. The increase will be around 60–70kg of CO2 per tonne of reformer feed for the two units considered, thus increasing the total carbon emissions related to reformer furnace firing to 205 and 230kg of CO2 per tonne of feed respectively.