How to Measure Total Concentration Indirectly

Hi Vulcan,

In the example you gave, if 4,000 liters is the entire flow it cannot be all water. The coffee must have some volume.

EDIT: I see PWSlack posted a similar comment while I was composing.

Weigh a given volume of the liquor.

Boil off the water.

Weigh the residue.

Divide.

As PWSlack has said you make the wrong assumption in that flow volume is 100% water. I would also like to add that also the dried coffee solids has retained some of the water. There will be some molecular bond of the water molecules to the coffee solids that are not removed in the drying process. Moisture content is between 3-3.5% usually in coffee spray driers. So some of your solids weight is actually water.

First question is- does all of the coffee dissolve or is there any particulate matter? That can make a huge difference.

In your example, that formula is useless. The difference between mass and volumetric flows simply reflects the density of the liquor, as the combination of water and dissolved coffee together total 4,000L AND 4,500 kg. You may be able to work something out though:

1. Some coriolis meters can read/calculate temperature, density, mass flow and volumetric flow. If your unit can do these it will make life much easier.
2. Find out what the specific gravity of your coffee solution is over the specific temperature and concentration ranges encountered in your process (this may best be done by a series of known concentrations, checking each over the temperature range). Note that all of this is based on the liquor being filtered clear of particulate, and of a consistent formula. If different blends have different specific gravities, you may have to chart for each one. As a suggestion, you might find that the density/concentration ratio is non-linear but the temperature difference is, meaning that you would need a table for the values but not the temperature as the result could be divided by a constant and then referenced to the table to derive the concentration.
3. Calculate density and temperature for each batch, and reference to your standards to determine the concentration.

Take a watch glass and weigh its tare weight. Take a known volume or close to it and place the coffee solution in it in a watch glass. That is your gross weight. Place the watch glass either on a steam bath, in an electric heater of the appropriate size, in an infrared heater box or best of all a microwave oven set on the low or medium power range. Heat the solution until all water has been driven off. Measure the weight of the watch glass. This is the net weight. Gross - tare = material start weight. (Net weight)/(material start weight) * 100% = per cent solids. Using the microwave or the infrared units are the fastest.

Good Luck, Old Salt

You need to manufacture some simple samples from your dried product, simulating the process in reverse and get a chart of concentration versus density.

You already know that 60% solids weighs in at 4500/4000 kg/L (or 1.125). If you make a solution at half that you should get a prediction line for your liquor and then check for linearity by the third already known point that a 0% concentration would have SG of 1.000.

If the prediction line passes through 1.000 for 0% concentration then you are reasonably sure you have a linear relationship.

For alternate methods, consult the sugar industry. They would have developed a method years ago to determine yield from cane syrup. Note: 2kg of sugar disolved into 2kg of water makes slightly greater than 3L of syrup (Used to feed my bees in late winter if they're running out of honey to consume.)

'Really appreciate the answers received so far. I'll try to comment on everyone's comments.

I'd like to avoid taking samples and making offline analysis (refractometer, oven methods, or others) because the operators sometimes forget to take regular samples, especially during the night shift. If I could get the Tc from the available instruments that I have (mass flowmeter, positive displacement pump, variable speed drive, temperature, and pressure), that would be great. The result doesn't have to be very accurate (+/-3 or 5% would be fine).

I kinda failed to consider the moisture content of the coffee powder after spray drying (thanks ozzb ).

I like JNB's suggestion of taking several samples and building a table or graph of density vs concentration or specific gravity vs concentration. That data can then be used to come up with a formula. Our coriolis flowmeter only has one output, unfortunately. Time to replace it I guess.

Thanks guys!

regards,

Vulcan

Density is an intensive property that does not scale directly with per cent solids.

What you will have to do in this case, since coffee is not a single substance, but is a complex mixture of compounds, and since coffee extract is not all in solution as you state, but part of it is simply emulsified with coffee oils and in suspended in the water.

To have any chance of metering the concentration, you could try a brix meter (refractive index), if you can get light through the extract at all, or use an online pyncnometer (measures specific gravity/density of the extract), or a calibrated specific gravity sond that floats in a vertical column of the extract, or you can use a calibrated volume (volumetric flask), and laboratory scales. To calibrate, you will need extracts of varying strengths starting with the strongest. Record the s.g. and/or density of each extract, then evaporate away the water at a controlled temperature, and weigh the solids. You must know the weight of the original extract as well that was subjected to evaporation.

Once you have calibrated and are satisfied, then you have a direct conversion from density to % solids in extract.

Another possibility that I have no experience with but understand, is the use of online Near Infrared spectroscopy. This has the potential of revealing more information than just % solids, and you still have to calibrate a method. Another method is simple visible light colorimetry on short optical path, since the OD of the extract is so high, otherwise you get no signal. I recommend a cheap spectrophotometer tuned to a green wavelength of 560 nm, calibrated, also try 700 nm (red end of visible light), and also try one around 450 nm in the blue end of the spectrum. The three values may allow you to make much more precise assessments of the quality of the roast.