Distilla'Sun (English version)

Distilled waters obtained by steam distillation are widely used in Morocco, examples being orange blossom water in cooking, rose water to remove make up, as well as thyme and rosemary water for their curative virtues for respiratory and digestive problems. Our Moroccan partners, with whom our school has maintained a partnership since 2012, asked us to design a system for steam distillation using solar energy, therefore limiting both the emission of greenhouse effect gas and deforestation while also making it accessible to as many as possible.

At first, basing ourselves on the traditional system, we had to think out a device allowing us to distill and optimize the performance and the cost of it. Then, we had to verify that the waters we obtained were close in both quantity and quality to those obtained using the traditional method.


Mimicry between traditional method and solar method

Écrit par Rohan NOWAK et Waldan GIRARD le . Publié dans How to distill using solar energy ?

We went to Morocco from 18 to 20 November 2015 to meet artisans and women from the Medina who distilled floral waters. We realized that this tradition persists in certain families and that some artisans sell their products in the souk. We gathered quite a lot of information relative to the secrets concerning the process and the ratio of flowers/water used/distillate obtained.

The principle is simple and close to the one we encountered in a physics and chemistry lesson, called hydro distillation:

Œ       Water is brought to a boil

       The vapors pass through the flowers from which we wish to extract the essential oils

Ž       The oils are then carried by the water vapor in which they are entirely soluble in gas form

       These vapors are liquefied in a cooling system containing cold water 

       We recuperate a distillate in which it is sometimes possible to clearly distinguish an organic phase floating if the oils are really not very miscible in water in the liquid state, as we could see with rosemary.



Table 1: Choice of materials based on advice from Moroccan artisans

Advice from “Moroccan specialists”

Choice of material

Limit the loss of vapor

Airtight container: pressure cooker with a seal in good condition

A sufficient power source for good vapor output

Source providing the most power: “Solar dish” concentrating the sun rays under the pressure cooker

Maintain the plants as dry as possible

Paint the exterior black so the whole pressure cooker heats up and not just the underneath.

5L of water for 1 kg of flowers

In an 8L pressure cooker, only 2.5 L of water and 500g of flowers to leave a space between the water and the flowers

All the vapor must go through the plants

Basket of the same diameter as the inside of the pressure cooker to limit the amount of vapor that doesn’t pass through the plants

Maintain the distillate cold with a damp piece of cloth: their cooling system liquefies the vapors but only cools the distillate a little bit

Preference for a bigger cooling system such as in Fig. 1 so as the waters recuperated are cooler to limit the loss by evaporation, in the absence of a real Florentine vase which would be too expensive

Preference for a glass bottle

Glass bottle to recuperate the distillate. Test made with a plastic bottle: the non-polar oils adhere to the non-polar polyethylene


The choice of a parabolic dish

Écrit par Rohan NOWAK et Waldan GIRARD le . Publié dans How to distill using solar energy ?

We realized we would be obliged to use a parabolic dish instead of a solar oven when we saw how fast the artisan we met got the water to boil. Our predecessors had tried to adapt their solar oven for distilling, but found that the heat transferred to the water wasn’t sufficient to bring the water to a rapid boil. On top of that, the time and cost of construction of such a system were too great. We decided to orient out research towards the solar dish and recycling satellite dishes.

a)Average power provided by a gas burner

We evaluated the heat flow transferred to water by a gas burner to compare it to that from a solar dish, the goal being to choose the size of the satellite dishes we would recycle, since their diameter varies between 50cm and 2m.


Experiment: Measure the heat flow provided by a traditional gas burner


-              Fill pressure cooker up with 6L of room temperature water

-              Heat using gas burner on full power

-              Time how long the water takes to boil


Observations: The water boils in 17min

That leads to:  

With cwater = 4.18 J.g-1.K-1, the calorific value of water per unit of mass

That’s why  

Conclusion: This heat flow is high, compared to the solar constant or solar flux above the earth’s atmosphere which is 1361 W.m-2. The solar flux on the ground will necessarily be lower so even with an efficiency of 100%, it seems that a dish with the biggest diameter as possible is necessary.

b) Power provided by the solar dish

The purchased solar dish we have in Tours and in Marrakech has a diameter of 1.50m. The heat flow of a gas burner doesn’t vary very much between different trials whereas the heat flow from a solar dish varies according to the sunshine. Therefore we tried to determine the efficiency of our device for a certain amount of sunshine, assuming the efficiency of heat transfer is independent of the amount of sunlight it receives.


-          Place the same pressure cooker containing 6.0L of water on the central support of the solar dish and close it while measuring the temperature of the water

-          Orient the dish so as it is facing the sun on the horizontal axis and adjust it vertically so as the patch of light where the sun rays converge is under the pressure cooker

-          Measure the angle of incidence of the sun rays and the inclination of the dish with respect to the horizontal axis

-          Time how long the water takes to boil


Measurements taken on 21 Jan 2016 at 11:15:

Dt = 52 min / qi= 20°C / qf = 100°C

Angle of incidence of the sun rays arays=33° and inclination of the dish adish=44°



The power provided by a solar dish in January is considerably lower than that from a gas burner. However, January isn’t the period during which the Moroccans distill their floral waters; they distill during flowering time, which is from April to June. So as to estimate how long the water would take to boil, or how much heat is transferred to the water, in that period of the year we evaluated the efficiency of our device under the experimental conditions in January.

Thanks to an internet site, we were able to get the maximum solar flux  at ground level according to the day of the year. On the 21 Jan at 12:00 the solar flux was of 681 W/m² and at its highest the sun was 38°.


It’s important to note that the sun rays don’t hit the dish parallel to its optical axis, but with a 13° inclination (b=90-adish-arays).


So the total solar surface recuperated is decreased: S’=Scosb=0,97S. We only lose 3% of the total surface, and that allowed us to make our device more stable, as will be explained later.






Therefore the solar power recuperated by the dish is:


A.N.:  = 1172 W = 1.17 kW.


Previously, we discovered that the water recuperated 0.64 kW. Therefore, the efficiency of our device is 55%. What happened to the missing 0.53 kW?

First hypothesis: The first hypothesis we made was about the mirror paper: it absorbs a certain percentage of the light received. This coefficient of absorption probably also depends on the wavelength of the light that is used.


-          Measure the incoming power of an HE-Ne laser (633 nm) 

-          Measure the power reflected by the mirror paper for different angles


Conclusion: The coefficient of reflection (R) is 0.90 for a 10° angle


Second hypothesis: The pressure cooker doesn’t absorb all the light received, even when painted black.


Experiment: Using a radiometer, we measured its albedo, which means the percentage of light sent back by light scattering.


Conclusion: 9% (±1%) of the light is scattered, so the pressure cooker only absorbs (91±1)% of the light received, its coefficient of absorption is a= 0,91.


Therefore, we have explained that regardless of the incident flow on the dish, the power absorbed by the pressure cooker is Pabs = aRPsolar = 0.82 Psolar. That explains a loss of 18% but not of 49%.


Third hypothesis: The pressure cooker, when heated, will lose energy by radiation, and also by convection and conduction, because it isn’t at the same temperature as its environment.

The diagram below sums up all the heat transfers that take place:


Knowing these different losses, we wanted to establish a model enabling us to evaluate the time that is required to bring a certain volume of water to a boil with a given amount of sunshine. We won’t take into account the losses by conduction as the contact between the stand and the pressure cooker is small (even if they might not be negligible).


- Loss of heat due to thermal radiation 

- ε is the emissivity of the stainless steel covered in black. With the hypothesis of the black body, ε=α=0.91 

- σ = 5.6696.10-8 is the Boltzmann constant

- Tcooker is the temperature at the surface of the metal (in K)

- s is the surface area for heat exchange between the pressure cooker and the air (in m²)

It is important to note that these losses depend on the temperature, which varies.


- Loss of heat due to natural convection

- s is the surface area for heat exchange between the pressure cooker and the air (in m²)

- Tair is the temperature of the surrounding air (in K)

- h is the heat exchange coefficient of the air. It depends on the temperature, the viscosity, the conductivity, the diffusivity, the density of the air… etc. In our case, we took the maximal value of h (10 W/m².K) that we found on internet so that we don’t minimize the losses.

Once again, these losses depend on the temperature of the pressure cooker, which varies.


If we only consider these losses, the conservation of energy can be expressed as follows:

Pwater/cocotte= PABS – (PRAD + PCONV)


This leads to a nonlinear first order differential equation. We solved it using the Euler method.

Knowing the initial temperature of the water, bit by bit we can determine the time taken to bring the water to a boil using a small increment operator


Table 2: Theoretical estimation of the time to bring water to a boil


t (in s)

t (in min)

T ( in K)




















q (in °)






D (in m)






d (in m)






H (in m)






S (in m²)






s' (in m²)






Tout (in K)






dt (in s)






m (in g)






h (W/m².K)






F (W/m²)






Pabs (in W)






mAl (in g)






caluminium  (J.g-1.K-1)














Using this technique and our hypotheses, we determined that the water should have boiled in 48 min and not 52min, which is a difference of about 8%. When we started the experiment at 11:00am the solar flux wasn’t at its maximum since the height of the sun was 33° and not 38°, which was the theoretical maximum height for that day. We didn’t take into account the losses by conduction or by forced convection (wind)… Therefore there is room for improvement of our model.


Thus for our subsequent trips to Morocco we had available two different methods for estimating by the time it would take to bring 20°C water to a boil:

-          Either by assuming that the efficiency will always be the same (50%). Since the instantaneous losses are proportional to the temperature of the water or to its temperature to the power of 4 and not to the solar flux, the efficiency will increase with increasing sunshine.

-          Or by using the model above.


Table 3: Theoretical estimation of the power of the parabolic dish and heating times



End of March

Beginning of May

Solar flux (W/m²)



Solar power (kW)



Average heat transferred to the water  (kW)


1.03 (50% compared to a gas burner)

Heating time (min) for 2.5L* for an efficiency of 55%



Heating time for 1.5L* with the theoretical model




Distillation time for 0.75L* (min)



: Quantities used for our distillations.


Calculation of the time taken to heat water from 20 to 100°C:   

Calculation of the time of distillation:  with Lvap,  = 2257 J/g   


In conclusion, we can note that in the season during which the solar dish is destined to be used, it only provides 50% of the power that a gas burner provides. Nevertheless, our system showed potential because it uses a renewable source of energy that is free and clean of greenhouse gases. 

Hosepipe for the vapors

Écrit par Rohan NOWAK et Waldan GIRARD le . Publié dans Optimization of the device after the first tests and discussions with specialists

When meeting with distillation specialists, Dr Jalil Belkamel, a Moroccan aromatherapist who is a specialist in plant extractions, and also Philippe Bertrand, a consultant on plant extraction processes, we clearly noticed their reluctance about the use of a hosepipe (used for in bathroom showers) because the internal surface is covered with rubber which is probably non-food grade. With a high temperature, the rubber deteriorates, solidifies and cracks. In this way, particles from this rubber could pollute the oils.


Fig. 20: Cracks formed in the rubber                         Fig. 21: Inside the hosepipe


We immediately decided to sacrifice one of these hoses to analyze its composition.

Firstly, the inside of the hosepipe did seem to be made of rubber. The problem is the rubbers used are synthetic and so they probably comprise some compounds added during the synthesis such as phthalates or other molecules to increase their flexibility or their lifetime.


Fig. 22: Skeletal formula of a phthalate 


Hypothesis: The oils contain impurities because of the rubber

Experiment: To check for potential impurities in the distilled waters, we decided to do 3 extractions on distillates obtained using 3 different hosepipes (made of 3 different materials) having the same diameter (10mm) and under the same distillation conditions (same heating power from a gas burner, same plants (dry rosemary) and same volume of perfumed waters obtained). The pipes were:

-Annealed copper;

-Shower hosepipe

-Multilayer food-grade hosepipe made of several layers of polyethylene;


We analyzed the floating oils by GC, but this time, we did it more slowly and followed it by mass spectrometry.



Black: Multilayer polyethylene

Pink: slow distillation (1h)

Blue: Hosepipe

Brown: Annealed copper


Fig. 23: Chromatogram of the oils obtained with different pipes

Conclusion: We did not detect any difference in the oils obtained. However it is not improbable for phthalates to be present in the oils, but in too much weak concentrations to be detected by the technique used. The shower hosepipe used for this experiment had not been used a lot and it is only after a series of heating and cooling cycles that it deteriorates. So, it is possible to use a copper pipe as Moroccans do but the device will lose mobility. Therefore we would need to put the dish+the condenser on the same rotating plate.


Vapor outlet

Écrit par Rohan NOWAK et Waldan GIRARD le . Publié dans Optimization of the device after the first tests and discussions with specialists

During the previous experiment, we were surprised by the fact that we obtained a layer of what looked like dust between the aqueous phase and the oils. An effect of the hosepipe was quickly eliminated because all the pipes lead to the same quantity of dust.

Hypothesis: According to Philippe Bertrand, the dust could be the consequence of the output being too high. He told us again that a good distillation had to last about one hour and not only 30 minutes or less as we did distilling with the gas burner.

Experiment: We immediately repeated the previous experiment, reducing the heating power provided by the gas burner to the water.




Fig. 24: Distillate from

1) High power heating, distillation in 20min

2) Low power heating power, 60min distillation


Observation/conclusion : Le résultat est probant : la couche de « poussière » est nettement moins importante ! En effet, le débit des vapeurs étant moins important, elles n’entraînent pas autant de particules avec elles.

Observations/Conclusion: The result was compelling: the layer of “dust” is clearly less significant! Indeed, when the vapor output was lower less particles were carried with the vapor..

A second comment from M. Bertrand and M. Belkamel concerned the vapor outlet of the pressure cooker which, according to them, was too narrow. Indeed, when we observed the traditional still, the vaporoutlet narrowed progressively (gooseneck).

This allows:

-a reduction in the velocity of the vapors at the outlet and also

-avoidance of excessive pressure, and at the same time a rise in temperature in accordance with the perfect gas law. It is worth noticing that this elevation of the temperature, even if only slight, could facilitate some oxidation reactions of the oils, commonly called a Maillard reaction (caramelization is a Maillard reaction).

Experiment: We decided to enlarge the outlet hole for the vapors to see whether or not there would be a visual change in the quality of the oils. The best would be to have a hosepipe which narrows progressively like a gooseneck but our aim was to make this device affordable to everybody, so we made the progressive reduction by using fittings. We decided to use an outlet with a diameter of 16mm instead of the 10mm outlet we used beforehand. This increases the cross-sectional area of the vapor outlet by 2.6 (s’/s = R’²/R² ) and therefore, according to the Venturi effect, divides the velocity of the vapor at the outlet by the same factor.


Results: The temperature recordings indicated that under the same distillation conditions, there was an average decrease in temperature of only 0.8° compared to the 10mm outlet.

Conclusion: This small difference in temperature cannot produce caramelization of the oils and cannot denature the oils.


Fig. 25: Enlarged vapor outlet

Cooling the water

Écrit par Rohan NOWAK et Waldan GIRARD le . Publié dans Optimization of the device after the first tests and discussions with specialists

For the condenser, it is best to have the largest surface area possible for heat exchange between the copper coil and the cold water. The perfumed waters have to cooled as much as possible before they exit from the condenser because the higher the temperature is, more volatile the oils are so we will lose their olfactory properties (the saturation vapor pressure increases with temperature). We did tests with a coil cut to half the length but this obliged us to change the cooling water two times during distillation.

With 10L of cold water, we wanted to determine what final temperature the water inside the condenser would reach at the end of distillation. The water in the condenser is used to liquefy the vapors and cool the distillate obtained. To do the calculation we are going to make some simplifying hypotheses.

- The perfumed water will exit the condenser at a constant temperature, whereas in fact it will be colder at the beginning of distillation than at the end;

- All the water in the condenser/distillate is in a closed system, meaning that there are no external exchanges;

-The distillate is purely water;



- qi = 20C, the initial temperature of the water in the condenser;

- qf, the final temperature of the water of the condenser;

- Vwater = 10 L the water volume in the condenser tank ;

- Vhydrolat = 0.75 L, the volume of the hydrolat (=distillate) obtained;

- Lliq = - Lvap = -2250 kJ.kg-1, latent heat of condensation for water ;

- cwater = 4,18 kJ.kg-1.°C-1 specific heat capacity for water

If we neglect the exchanges with the outside:

DHwater + DHhydrolat = 0

With DHwater = mwatercwater(qf-qi)

HDhydrolat = -Lvap.mhydrolat + mhydrolatcwater(qf-100).

So : mwatercwater(qf-qi) -Lvap.mhydrolat + mhydrolatcwater(qf-100) = 0

qf =   so qf = 63°C

This is why we have to change the water during the distillation, because at the end, the temperature will be too high.

We tried to find a way to cool the water in the condenser. In electric power plants, water drops on grids or fans are used to cool the water but for us it is impossible to do that. However, we thought about a way using evaporation, as is used in the desert to keep drinks cold. We wanted to see if having a damp cloth around the condenser would have an efficient effect. If a damp cloth surrounds the tank, the water in the cloth will evaporate and since the vaporization is an endothermal phenomenon it will absorb heat energy from the condenser.

The factors favoring vaporization are:

- increasing the surface area exposed (spreading the cloth favors vaporization) : we have to cover the whole tank with a damp cloth

- the type of wet material used (wool, cotton, nylon...) : what is the best for our use?

- the wind : windier it is, more efficient it will be, but in Morocco, it is not really the case... → forced convection

- heat input such as the sun favors drying by vaporization because it is an endothermic phenomenon, which means that it will take heat from its environment. So we should use a white cloth to avoid the drying by absorption of the solar radiation and keep the condenser out of the sun, in the shade of the dish. The cloth has to dry by absorption of the energy from the water and not thanks to the sun effect.

- The lower the relative humidity of the air the higher the rate of vaporization will be. This is one of the advantages in Morocco where the climate is dry.

During our previous trip to Morocco, we tested this hypothesis. At the end of the experiment, the temperature of the water inside the condenser was 51°C. Indeed, we noticed that the damp cloth used to surround the condenser dried quickly, but it was generally helped by the wind which was quite strong that day. We hope to repeat this experiment during our next trip to confirm or refute this hypothesis, by doing two distillations in parallel.