UF-201 and NF-201

The membrane section of the process has the aim of increasing the product purity in polyphenols and to reduce the amount of solvent in the product. Choosing a membrane with a molecular cut-off lower than the average molecular weight of the cellulosic material allows to obtain a retentate with low cellulose content and richer in the products of interest. Membrane separation is a unit operation that takes advantage of the trans-membrane pressure as its driving force, which allows the solvent to move through a semi-permeable membrane, along with the solutes whose molecular weight is below the cut-ff of the membrane. Membranes are porous, and it is actually through these porosities that the solvent and the light solutes are able to permeate. It’s usually considered a good practice to select a membrane whose pores have size about one tenth of the molecule to be removed

93 (Cheryan, 1998). The membrane filtration is structured in two steps: an

ultrafiltration step (cut off 1500 Da) and a nano-filtration (cut-ff 200 Da). The permeate from the UF is fed to the NF, from which in turn a polyphenols-rich retentate is obtained, which will in turn be further concentrated with an

evaporator and a drum dryer. Membranes are an interesting choice for their low encumbrance, availability on the market, low operative and chemicals cost. The choice of the working conditions of the membrane is based on technical and econimical considerations that will be explained in this paragraph. The feed pressure is instead chosen according to the manufacturer reccomendations: the pressure is set to be about 80% of the maximum allowable value of the module.

To start with the sizing of the membranes, it’s necessary to fix a ratio between permeate and feed flowrates, or alternatively, a Volume Concentration Ratio; in both cases, they allow us to calculate the flowrate of either permeate or retentate based on the volumetric flowrate of the feed. This parameter is a desinger’s choice. It is necessary to know the rejection coefficients of the various solutes to caclulate their concentration in the product streams. The rejection coefficient is a value beteween 0 and 100%, the higher the more pronunciated is the tendency of the solute to stay in the retentate. The concentrations for solutes in the retentate are given by:

𝐶𝐴,𝑟𝑒𝑡𝑒𝑛𝑡𝑎𝑡𝑒 = (1 −𝑄_{𝑝𝑒𝑟𝑚𝑒𝑎𝑡𝑒}

𝑄_{𝑓𝑒𝑒𝑑} )^−𝑅𝐴 (6.8) Where:

• 𝐶𝐴,𝑟𝑒𝑡𝑒𝑛𝑡𝑎𝑡𝑒 is the concentration of solute A in the retentate stream

• 𝑄_{𝑝𝑒𝑟𝑚𝑒𝑎𝑡𝑒} is the volumetric flowrate of the permeate

• 𝑄_{𝑓𝑒𝑒𝑑} is the volumetric flowrate of the feed

• 𝑅_𝐴 is the rejection coefficient of solute A

With this information it’s possible to determine the mass balances over the membranes. The ratio between the feed and the permeate allows to calculate the solvent flowrate in these two currents: if density variations with the solutes’

concentration is neglected, this also allows to calculate the retentate solvent flowrate by difference. Knowing the solvent flowrate in the retentate and the solutes’ concentration in the same current, it’s possible to close the mass balances for every current.

𝑄_{𝑓𝑒𝑒𝑑}𝐶_{𝐴,𝑓𝑒𝑒𝑑}= 𝑄_{𝑝𝑒𝑟𝑚𝑒𝑎𝑡𝑒}𝐶_{𝐴,𝑝𝑒𝑟𝑚𝑒𝑎𝑡𝑒} + 𝑄_{𝑟𝑒𝑡𝑒𝑛𝑡𝑎𝑡𝑒}𝐶𝐴,𝑟𝑒𝑡𝑒𝑛𝑡𝑎𝑡𝑒

To calculate the minimum area necessary to achieve the desired permeate and retentate flow, it is necessary to have data on the permeability of the membrane, that is to say, the permeate volumetric flowrate passing through the unit area of membrane in the unit time, per unit of driving force (trans-membrane pressure).

𝐴 = 𝑄_{𝑓𝐸𝐸𝐷}

𝜑 𝑇𝑀𝑃 (6.9)

94 Where:

• 𝜑 is the solvent permeability

• 𝑇𝑀𝑃 is the trans-membrane pressure

Knowing the amount of driving force and the required volumetric flowrate of permeate, the minimum area of the membrane has been calculated. This area needs to be further increased; membranes accumulate fouling the hinder their performance, to it’s necessary to proceeed with periodical cleaning of the

membranes modules. For this reason, the number of real modules is higher than the ideal one. In this way, the feed can be treated without interruptions even when some of the modules require to stop. A module enters the cleaning phase when a decrease of more than 20% of the nomial permeate flowrate is registered.

At this point the module is isolated from the rest of the rack my means of shut-off valves, and starts the cleaning routine. The cleaning routines are usually conducted as per reccomendation of the manufacturer, and incluedes clean water cleanings, alternated with a caustic and a acid cleaning. The solutions have been surmised to be able to remove a 0.01 kg/m2 of fooling in the case of the basic solution and 0.005 kg/m2 in the case of the acid solution, for each cleaning. After the cleaning, the solutions are returned to their storage vessel to be used again.

The solution can accumulate a certain amount of fouling before becoming too dirty to be used; at this point they are discharged. Using a maximum tolerable solid concentration of 5 kg/m3, it’s possible to surmise the number of cleaning cycles a solution can undergo. An operational time has also been surmised. To do so, it is necessary to obtain data on the fouling behaviour of the membrane. This data is usually presented in the literature as filtrate flowrate as a function of time. The logarithm of the ratio between initial permeate flux and permeate flux at a generic time is a linear function, whose angular coefficient is called fouling factor. This allows to estimate the time at which the permeate flux will decrease to 80% of the initial value. Regarding the best set of conditions for the

membranes, the procedure followed has beeen the following. For both the UF and NF steps, several combination of values of the ratio 𝑄𝑝𝑒𝑟𝑚𝑒𝑎𝑡𝑒

𝑄𝑓𝑒𝑒𝑑

⁄ for both membrane steps have been evaluated.. For each combination, the profitability of the section has been evaluated based on profits and costs:

• The revenue of the product sales. This has been calculated as the amount of the polyphenolic-rich material obtained (considering 10% final moisture of the product) times the price of the materiale per kg. This in turn has been estimated as a linear function of the product purity in polyphenols, with a price of 50 euro/kg for a product with 20% purity and of 200 euros/kg for a product with 90% purity.

• The annualised cost of the membranes, accounting for 200 euros for square metre of memrbane

95

• The annual cost for the chemicals necessary for the membranes’

regeneration

• The annual cost for the steam needed to evaporate the residual solvent

• The annualised cost of the multiple evaporators

This allows to surmise how the profitability of the purification section varies with the 𝑄_{𝑝𝑒𝑟𝑚𝑒𝑎𝑡𝑒}

𝑄_{𝑓𝑒𝑒𝑑}

⁄ ratio, and to select the most profitable one, if other technical consideration do not prevent it. In this case, low values of the ratios give higher profitability, but the lowest ones give a finale product with a concentration of cellulosic material too high to be deemed acceptable. On the other hand, too high values would cause the liquid fed to the evaporators to have a high solid content that prevents the evaporators from properly functioning. The selected membranes are from the Desal GK manufacturer. They are spiral wound-membranes, easily assembled in racks.