Comparing Knock Engines with NIR Analyzers for In-Line Blending

By Ara Barsamian


With the ultra low sulfur/ethanol gasoline blends around the corner, one would like to get the blends done right the first time around, preferably being capable to certify while blending to a ship or pipeline.  Currently, the officially sanctioned ASTM method for in-line blend certification of gasoline octanes use CFR knock engines (ASTM D2885); however, these are very maintenance intensive, temperamental, and very difficult to use for routing blending.

In contrast, NIR, NMR or Raman-based analyzer technologies are very easy to set up and run.  In addition, they can measure additional properties, thus saving the cost of additional 7 to 10 analyzers for other properties.  Their "hassles" include the effort to get "special waivers" to be able to use in the D2885 comparator mode for blend certification, and development of robust property prediction models.

Because of the great interest in the subject, a comparison was made between the Life Cycle Costs of a CFR knock engine installation against a NIR multi-property analyzer.  The NIR type analyzers are generally the most cost-effective, and if the modeling is right, very reliable and hassle free.

Comparing CFR Knock Engines with NIR Analyzers

We want to compare two things:  life-cycle costs, and ease of use.

Life-Cycle Costs

There are four cost areas considered:

The bottom line is that NIR (or NMR<Raman) runs circles around a knock engine, with NIR running at about 3.4M$/10 years vs. 5.9M$/10 years for the knock engines.  With the knock engines, we need to also add the cost of another five to eight analyzers (e.g., RVP analyzer, distillation analyzer, density analyzer, and aromatics, benzene, olefins and oxygenate analyzers) together with their sampling and sample conditioning systems, protofuel check systems, etc.

Ease of Use

The CFR knock engines require about 24 discrete steps to start up the engine, warm it up, and make it ready to analyze the blended stream; conversely, it requires an additional ten steps to shut it down.  Just to give an idea of the typical steps, the following is an abbreviated list for the RON engine:

And this assumes you don't have plugged filters, or N2 blanket in prototanks leaking into the fuel and choking the engine, etc.

Now, this whole sequence has to be done for the second (MON) engine, etc.  The start up and shut down can be partially automated with PLC's, but it still requires a technician or blender to keep an eye on it.

For NIR, the whole start up and shutdown sequence is controlled by the DCS.  Since there are no mechanical components, the sequence includes optical and electronic self-diagnostics per ASTM D6122 to make sure it works properly, and then it is checked against the correct protofuel (switched in automatically by the DCS software).  If it passes both tests, then it is automatically brought on-line.  It typically provides blend line readings every one to two minutes (of all 10-15 properties, not just octanes), depending on design.  In addition, every one to two minutes, there are performance monitoring algorithms checking the raw readings against predicted calculations, and the difference has to be less than the ASTM tolerance.