Oct-2015

Minimising gasoline specification 
give-away

Minimising the difference between a measured physical property and a critical specification for gasoline can have a significant impact on refinery profitability.

DAVID SEIVER
Valero

Article Summary

In 2014, roughly 136.8 billion gallons (or 3.26 billion barrels) of gasoline were consumed in the US, a daily average of about 374.7 million gallons (or 8.9 million barrels). This was about 4% less than the record high of around 142.5 billion gallons (or 3.4 billion barrels) consumed in 2007. Gasoline is one of the major fuels consumed in the United States, and it is the main product refined from crude oil. About 10% of the volume of finished motor gasoline now consumed in the US is ethanol. Diesel production has outpaced gasoline production in the last five years, but still gasoline production growth from 2009-2014 was up by 376 000 barrels according to the US Energy Information Administration (EIA). In 2013, gasoline (including fuel ethanol) accounted for about 61% of all the energy used for transportation, 45% of all petroleum consumption, and 17% of total US energy consumption. Although most refineries are shifting more towards distillate production, still nearly 45 barrels of gasoline are produced in US refineries from every 100 barrels of oil refined to make petroleum products. Even outside the US, where distillates are the fuel of choice in most countries, gasoline represents a significant percentage of the clean products produced in a refinery and thus optimising give-away reduction still represents a significant strategic opportunity.

Minimising the difference between a measured physical property and a critical specification (‘give-away’) on a tactical level can have a significant impact on refinery profitability. In a highly competitive, low margin, fungible commodity market like that of clean petroleum fuel products, this could mean the difference between keeping a refinery running or shutting it down. Refinery blending typically consists of gasoline and diesel product blending and can be considered the ‘cash register’ of the refinery. It is the last chance to optimise composition and to get as close to product specification as possible without excessive give-away. If there is specification give-away at blending, it is truly lost revenue and cancels out benefits gained in upstream unit process areas. At a typical refinery, optimised gasoline blending could represent a majority of the site’s total advanced process control (APC) savings, and yield in excess of $20 million/year in bottom line savings. Bear in mind that a small reduction in give-away yields impressive results through scale-up. Although with recent declines in crude oil prices, gasoline prices have also declined, refining margins have still held up and key gasoline specifications of interest like octane and Reid vapour pressure (RVP) have stayed relatively constant. (see Figure 1) This fact alone should buoy the long term robustness of tackling this APC strategy.

How much are we giving away?
Until we measure, we cannot control. So let us examine, or measure, how much potential give-away is out there, so we can ascertain the economic potential for reduction. First of all, what are the measurable key properties of gasoline that we can put a price to, and can potentially control? Two key properties of concern are octane, measured in octane number barrels or ON-barrels, and RVP, measured in pounds per square inch barrels or PSI-barrels. Generally all other constraining gasoline specifications like V/L, T50, and VOC reduction requirements for RBOBs can be converted to RVP equivalents. Often octane and RVP can be controlled somewhat independently of each other, but not always and thus advanced techniques that will be described later are needed. Conservative industry average estimates for octane and RVP specification give-away show at least 0.5 ON and 0.4 psi respectively per barrel of gasoline blended. Said another way, on average the pump road octane (R+M/2) for gasoline is 0.5 ON higher than specification, so that if you are buying 87-octane at the pump, you are often getting 87.5-octane gasoline. 

Of course, this varies significantly by site and a few caveats are due at this point. Currently in the US, very few refiners certify gasoline using on-line analysis. Many use on-line analysers to control, but ultimately certify the gasoline using ‘off-line’ lab results. Sites where on-line certification occurs will generally experience significantly less give-away than is addressed in this article. Also, the estimates above do not include ‘forced’ give-away for VOC reduction requirements and also exclude seasonal transition periods where excessive give-away occurs when ‘trimming’ terminal tankage to the new seasonal requirements. Also, some refineries may be limited by logistics (in other words, no way to export high value gasoline components above gasoline sales prices), or in some rare cases markets cannot increase premium sales in a tight/constrained market to take advantage of reduced give-away. For example, if reduced octane give-away led to increased high value blend component stocks like alkylate, but the site was already at the lowest alkylate production or yield and could not sell it on the open market above gasoline prices, the optimisation may not lead to bottom line savings for the refinery. In the writer’s experience, the above caveats are relatively rare and should not significantly alter the overall savings potential described herein.

As mentioned already, today’s US refineries produce roughly 8.9 million barrels of gasoline every day. Although the sales values of octane and RVP vary by region by season, over the last several years a good, conservative working average value for octane in the US is $1.20 ON-barrel, and $1.00 psi-
barrel for RVP or RVP equivalent (calculated generically by the differential in premium versus regular sales prices, and differentials in low cost components like butane versus gasoline). Performing the calculation yields total annual give-away in the US of about $2.0 billion for octane and $1.3 billion for RVP. Even if only 50% of this is recoverable, there is a potential for over $1.6 billion in combined give-away savings as an industry in the US alone.

Evidence that I am not alone in seeing these potential savings are small ‘butanising’ operations that have started to become prevalent in the US that take finished gasoline off major pipelines and ‘trim’ them closer to specification, capturing the yield increase using a cheap gasoline volume expansion component like butane. This appears to be a clear indictment of the refining industries’ moderate, if not poor, success in reducing give-away thus far. The economic case for pursuing gasoline blending optimisation thus seems to be compelling.

What does it take to tackle gasoline blending optimisation?
Many vendors will tell you otherwise, but evidenced by all the failed attempts scattered across the industry, gasoline blend optimisation is not a quick, cheap, or easy endeavour to tackle. This is true for several facts that are often glossed over, or poorly understood by those embarking on this for the first time. First of all, of all APC projects, this is one that involves intimate coordination and support by disparate refining groups who historically have not worked closely or well with one another. If you were to apply APC at the FCC of a refinery, likely only the operations group of that unit needs to be involved for success. For APC at blending, operations, lab, planning and economics, and analyser groups need to be engaged and working well together for the project to be successful and more importantly sustained long term. Ownership of the overall systems is crucial to get sorted out early in the project and a clear understanding of who is responsible for each component of the installed system is required. Routine meetings of all responsible parties after implementation to address on-going concerns and to review issues is also highly recommended, as well as setting up key metrics monitoring to set priorities and ensure management support going forward.

Another often poorly understood fact of the matter is that any and all upstream units related to gasoline can, and will, affect the performance of the gasoline blending optimiser. Changes in upstream unit operations will change blend component properties, and the associated interactions with the other blend components. So in other words, the further downstream the unit is, the more the upstream units will affect the results, and gasoline blending is of course at the end of the line. This becomes evident when making and maintaining robust, up-to-date, inferred property models 
for the on-line analysers, for instance spectroscopic analysers like FTIRs and Ramans, which are often used to control and reduce give-away.