To: D. DiBiasio, VP Process Development

From: M. Lundgren, M. Bodanza

Date: November 20, 2015

Subject: Cameroon Renewable Energy Project: Assessment of E85-Biofuel Production

Introduction

The African Centre for Renewable Energy and Sustainable Technology (ACREST) and French investor, Donalalé Boye, have hired our team to investigate the feasibility of high-grade ethanol

production from palm wine in a process facility located in Cameroon, Africa.1 E85, if refined properly, can heavily impact Cameroon’s transportation industry by providing efficient, clean-

burning and well-resourced biofuel, thus increasing energy availability.

Our objective will be exploring the practicality of ethanol-fuel production in Cameroon. ACREST has provided our team with design parameters and a product goal. With this, we must determine the feasibility of attaining these constraints using theoretical efficiencies and data from a pilot

fractionator. In addition, we must provide estimates for the amount of vehicles which benefit annually from the E85-Biofuel, and recommendations for safe and novel ways to repurpose and/or

recycle the waste water from the ethanol-water separation. Along with these considerations, we must prove this process is not only possible within the given design parameters, but is also economically reasonable.

Fig. 1. Pilot Column Diagram. Operation at total reflux.

Qc

Qr

V

L

𝑉ത

1

2

3

𝐿ത

11

For our pilot distillation column running at total reflux, interns from the Cameroon Academy of Petrochemical Engineering1 have collected data, for analysis, in order to find the information

(ethanol-water concentrations, efficiency, etc.) to design a full-scale column. This column will be presented, along with other data to ACREST and Donalalé Boye for use in the Mbouta process

facility.

Methodology

For a proper analysis on the given design column parameters and pilot plant, certain concepts must be used. For the pilot column analysis, constant molar overflow (CMO) needs to be valid. With

that in mind, energy balances can be done around the reboiler and condenser, and using data collected by the CAPE interns we can determine heat differentials. While realistically the condenser and reboiler are not adiabatic, the heat loss from either will be considered insignificant

for our purposes. With the heat duties on the reboiler and condenser known, we can find the molar vapor flow rates from the reboiler and top of the column through the appropriate relationship

between heat duty, flow rate, and enthalpy changes. Assuming the column operates at atmospheric pressure, an appropriate x-y diagram was used in conjunction with a McCabe-Thiele analysis to find the theoretical number of stages. With this result and the actual number of stages in the pilot

plant, the overall efficiency of the pilot column can be determined. This efficiency will be important later for the design column.

With the above data analyzed, attention must be turned to the design column investigation. As

before, a CMO validation must be done in order for us to make any further assumptions within the column. With this condition met, we can construct operating lines for the column. In order to complete this, several assumptions must be made which include the following: q-value and optimal

reflux ratio. The former was found after comparing a few different values and their effects on the design. The later was determined from standard operating conditions for most economical

conditions.2 With these assumptions McCabe-Thiele analysis was conducted on the VLE data to obtain the theoretical number of stages. From the efficiency obtained above, the actual number of stages was determined. This result was compared to the efficiency obtained from O’Connell’s

correlation

Additionally, ACREST has requested other information, including heat duties on the design fractionator, column diameter, and the amount of vehicles benefited from the given production

rate. The heat duties were found using independent energy balances around the condenser and reboiler. The column diameter was found utilizing the Souders-Brown equation. Finally, the amount of vehicles benefited was found by considering average car tank sizes.

Results and Discussion

Pilot Column Analysis

Refer to Appendix I (AI)

To begin our analysis, we first had to validate CMO conditions. First, the latent heats of water and ethanol were found to be within ~5% of each other. Then, the heat duties and heat loss were determined. The overall heat loss from the column was found to be 2.371 kW. As the column is

completely uninsulated and open to the atmosphere at the top of the column, CMO cannot strictly be applied here. However, for the sake of this column being a first pass at the operation, we

assumed CMO. The following tables and figures summarize these results and other required values of the column.

Percent difference of latent heats: 5.44%

QR QC QLoss

4.832 kW -2.461 kW -2.371 kW

Vapor flow rate at top of column Vapor flow rate from reboiler

174 mol/hr 404 mol/hr

Fig 2. McCabe-Thiele Analysis of VLE Data for Pilot Plant..

Column Efficiency: 36%

Design Column Analysis

Refer to Appendix II (AII)

Similar to the pilot column, CMO was validated using the comparison of the latent heats. However, for the design of this column, the column will be considered adiabatic, fully validating CMO.

As stated prior, many assumptions were made in the design of the column, chiefly the quality of the feed. Two values, 0.25 and 1.0, were of particular interest and individual McCabe-Thiele analyses were carried out. A quality of 0.25 was considered more economically practical, as it

required less stages. Additionally, the optimal reflux ratio was heavily assumed. From published guidelines,2 it was considered that 1.3*Rmin was a reasonable estimate. To allow for more

flexibility in operation, a reflux ratio above the optimal was used. These results are summarized below in the following figures and graphs as well as other pertinent values.

With the reflux ratio, an energy balance can now be done around the condenser to find Q c with

the vapor flow rate (V) and the enthalpy change. With Qc and the total energy balance of the

column, Qr can be calculated: 𝐹ℎ𝐹 + 𝑄𝐶 + 𝑄𝑅 = 𝐷ℎ𝐷 + 𝐵ℎ𝐷

However in order to prove the actual stage calculation is correct within one standard deviation, a

comparison must be made to industry standard column. The O’Connell Efficiency plot gives

different average efficiencies depending on the column’s temperature, relative volatility and

viscosity.

Table R.3

Above is the data calculated when finding the O’Connell correlation. It is clear the actual number

of stages found through the O’Connell correlation and the VLE diagram are very close (21 vs. 25

stages). This difference is within one standard deviation thus failing to disprove our Nactual

computation, making it statistically accurate for indursty use. The reason why the O’Connell

correlation number of stages is different then what was calculated is because the O’Connell

efficiency plot is made up of averages and does not consider other factors (such as feed quality)

which impacts the result.

With the validation of the design column’s internal schematics the diameter of the column was another valuable construction parameter to calculate. To do this the Souders-Brown Equation

was used:

V = k√ρl−ρv

ρv

This equation describes the relationship between the the relative densities of vapor and liquid in the column along with a constant k. From this V can be calculated and then used along with the volumetric flow rate to calculate the area at the top of the column.

Avg. Temp. Column (ºC)

EWavg mix (cP) O’Connell Correlation

NActual

85 4.956 0.3542 44% 21 Stages

A =vapor volumetric flow rate

V

With the area, the diameter can be calculated assuming a perfectly cylindrical column.

A =π

4d2

It was found that under the given parameters, a column with a diameter of around 0.338 meters is

required. If a higher q (1) was chosen at the beginning of the design column analyses, a larger

diameter would be expected because of the higher flow rates.

Finally, along with all the above data for the column design specifications, ACREST had

requested an estimate to how many vehicles would benefit annually from a 5,000,000 L/yr

product. An E85-Biofuel average fuel consumption and tank size was assumed in order to

calculate about 302 car-tanks/day or around 110,230 car-tanks/yr would benefit.

Figure 3. Design Column Diagram.

Percent difference of latent heats: 5.44%

QR QC QLoss

968,000 kJ/hr -4,270,000 kJ/hr 0 kJ/hr

Qc

Qr

V

L

𝑉ത

1

2

3

𝐿ത

25

D

B

Quality Theoretical Number of Stages Actual Number of Stages

1.0 15 41

0.25 9 25

Fig 4. McCabe-Thiele Analysis of VLE Data for Design Column. Feed quality of 0.25

Feed Line: y = −(0.25

0.75)x + (0.0891

0.75)

Rectifying Line: y = (0.8982)x + (1 − 0.8982) ∗ 0.8181

Column Diameter: 0.338 m

Cars that benefit from biofuel: 302 car tanks/day

Conclusions and Recommendations

The purpose of the proposed column is to provide Cameroon with a viable, new alternative energy supply that will hopefully benefit the economy of this restricted country. As the country faces strict

export policies concerning oil and petrol, a readily available and renewable energy supply is the advantage Cameroon needs to flourish and grow economically. However, the plant will have its

demands on the local area, and its benefit does not come without cost. It is important to further consider the impacts on both the current energy demands and environment.

While further investigation is necessary, our analysis provides hope for alternate fuel sources. It is our recommendation that further studying of design parameters be considered to optimize the

column to peek operation. Such parameters could include the conditions of the feed which have a high influence on the column as a whole in terms of its energy cost and capital cost.

Appendices

Appendix I: Pilot Column Analysis

Energy Balances

QR = −msteam∆Hsteam

QR = − (130g

min) (

1 kg

1000g)(−2,230.088

kJ

kg)(

1 min

60 s) = 4.832 kW

QCoolant = mH20 Cp,H20∆T

QCoolant = (1.4 gpm) (3785 cc

1 gal)(

1.00 g

1 cc)(4.184

J

g℃) (6.66℃) (

1 kJ

1000 J) (

1 min

60 s) = 2.461 kW

Qc = −Qcoolant = −2.461 kW

QLoss = −QC − QR = −2.371 kW

CMO Validation

𝑸𝑳𝒐𝒔𝒔 ≠ 𝟎

λH20 = 40.66 kJ

mol

λEtOH = 38.56 kJ

mol

λH2O−λEtOH

λEtOH= 5.44%

Assumed CMO to obtain an approximation.

Molar Vapor Flow Rate at Top of Column

QC = mV (hL − HV)

−2.217 kW = mV (−259.02kcal

kg) (

4.184 kJ

1 kcal)

mV = (0.002046 kg

s)(

3600 s

1 hr) = 7.36

kg

hr

nV = mV/MWavg

nV =7.36 kg

hr

42.42 kgkmol

× 1000𝑚𝑜𝑙

𝑘𝑚𝑜𝑙= 174

mol

hr

Molar Vapor Flow Rate from Reboiler

QR = mL ΔH

L

4.832 kW = mL (526.4kcal

kg) (

4.184 kJ

1 kcal)

mL = (0.002193

kg

s)(

3600 s

1 hr) = 7.90

kg

hr

nL = mL /MWavg

nL =

(7.90 kghr

)

(19.56 kgkmol

)× 1000

𝑚𝑜𝑙

𝑘𝑚𝑜𝑙= 404

mol

hr

Overall Efficiency

Eo =Ntheoretical

Nactual=

4

11= 36%

Theoretical equilibrium stages considered to be four rather than three because the reboiler is not a true equilibrium stage.

Appendix II: Design Column Analysis

CMO Validation

𝑸𝑳𝒐𝒔𝒔 = 𝟎

λH20 = 40.66 kJ

mol

λEtOH = 38.56 kJ

mol

λH2O−λEtOH

λEtOH= 5.44%

Overall Mass/Molar Balances

ρavg = wEtOHρEtOH + wH2OρH2O = (0.92 ∗ 0.789 g

mL) + (0.08 ∗ 0.9982

g

mL) = 0.8057

g

mL

Dm = ρavgDV = (0.8057 g

mL)(

1000 mL

1 L)(5 M

L

yr) (

1 yr

365 days) (

1 day

24 hrs)(

1 kg

1000 g) = 459.9

kg

hr

MWavg = wEtOHMWEtOH + wH2OMWH2O = (0.92 ∗ 46.07 g

mol) + (0.08 ∗ 18.02

g

mol) = 43.826

g

mol

Dn = Dm/MWavg = (459.9 kg

hr)/(43.826

kg

kmol) = 10.49

kmol

hr

xD =

wEtOHMWEtOH

[( wEtOHMWEtOH

) + ( wH2OMWH2O

)] =

0.92

46.07kg

kmol

[( 0.9246.07kg

kmol

) + ( 0.08

18.02kg

kmol

)]

= 0.8181

All other mole fractions can be found in a similar process.

F = D + B

Fz = DxD + BxB

F = (xD−xB

z−xB)D = (

0.8181−0.00394

0.0891−0.00394) (10.49

kmol

hr) = 100.3

kmol

hr

B = 89.81 kmol

hr

VLE Data and Operating Lines

Feed Line

y = − (q

1−q)x + (

z

1−q)

Assumed q = 0.25.

y = − (0.25

0.75)x + (

0.0891

0.75)

Reflux Ratio

Minimum: at pinch point

(L

V)

min= (

y2−y1

x2−x1)

Where the two points are on y=x line at xD and the y-intercept for the line that makes a pinch point with the VLE curve.

(L

V)

min= (

0.8181−0.105

0.8181−0) = 0.8717

Rmin = (L

D)

min=

L V⁄

1−L V⁄=

0.8717

1−.8717= 6.784

Optimal

Ropt = (1.1 − 1.4)Rmin = 1.3 ∗ 6.784 = 8.8192

Range of multipliers used in senior design textbook in Haley’s presentation.

(L

V)

opt=

Ropt

1+Ropt=

8.8192

1+8.8192= 0.8982

Rectifying Operating Line

y = (L

V) x + (1 −

L

V) xD

y = (0.8982)x + (1 − 0.8982) ∗ 0.8181

Stripping line specified by intersection of feed and rectifying line and bottom composition.

Overall Energy Balance

𝑉 = (𝑅𝑜𝑝𝑡 + 1)𝐷 = (8.8192 + 1) (459.9 𝑘𝑔

ℎ𝑟) = 4519

𝑘𝑔

ℎ𝑟

QC = V1(hD − H1) = (4519 𝑘𝑔

ℎ𝑟) (−225.64

𝑘𝑐𝑎𝑙

𝑘𝑔)(4.184

𝑘𝐽

𝑘𝑐𝑎𝑙) = −4,270,000

𝑘𝐽

ℎ𝑟

𝐹ℎ𝐹 + 𝑄𝐶 + 𝑄𝑅 = 𝐷ℎ𝐷 + 𝐵ℎ𝐷

𝑄𝑅 = 𝐷ℎ𝐷 + 𝐵ℎ𝐵 − 𝐹ℎ𝐹 − 𝑄𝐶

𝑄𝑅 = [(459.8𝑘𝑔

ℎ𝑟)(51.28

𝑘𝑐𝑎𝑙

𝑘𝑔) + (1742.4

𝑘𝑐𝑎𝑙

ℎ𝑟)(89.47

𝑘𝑐𝑎𝑙

𝑘𝑔) − (2202.2

𝑘𝑔

ℎ𝑟)(439.47

𝑘𝑐𝑎𝑙

𝑘𝑔)] [4.184

𝑘𝐽

𝑘𝑐𝑎𝑙] + 4,266,287

𝑘𝐽

ℎ𝑟

𝑄𝑅 = 968,000 𝑘𝐽

ℎ𝑟

Required Stages

Theoretically, 10 equilibrium contacts are needed – 9 column stages and 1 reboiler.

Pilot Efficiency

Eo =Ntheoretical

Nactual

Nactual =Ntheoretical

Eo=

9

0.36= 25 stages

O’Connell Efficiency

Tavg,column = 85℃

αEW,avg,column = 4.956

ln μmix = xEtOH,f ln μEtOH,85℃ + xH2O,f ln μH2O,80℃

ln μmix = (0.0891) ln(0.40) + (0.9109) ln(0.35)

μmix = 0.3542 cP

Note: viscosity of water is at 80ᵒC; this is highest available viscosity of water from table.

O’Connell Correlation:

Eo = 50.3(αμ)−0.226 = 50.3(4.956 ∗ 0.3542)−0.226 = 44%

Nactual =Ntheoretical

Eo=

9

0.44= 21 stages

Column Diameter

Souders-Brown Equation:

V = k√ρl−ρv

ρv

Where k = 0.107 m/s.

ρv =P∗MWavg,vapor

RT=

(1.01325 bar)(43.826g

mol)

(83.14bar∗ccmol∗K

)(358 K)(

1000000cc

1 m3 ) (1 kg

1000 g) = 1.4919

kg

m3

ρL =mEtOH+mH2O

mEtOH/ρEtOH+mh2O/ρH2O=

1 g + 99 g

1 g/0.789 gmL

+ 99 g/.992 gmL

(1000000 mL

1 m3 )(1 kg

1000 g) = 989.5

kg

m3

V = (0.107m

s) √

989.5−1.4919

1.4919= 2.766 m/s

A =vapor volumetric flow rate

V=

0.089887 m3/s

2.766 m/s= 0.08988 m2

A =π

4d2

d = √4A

π= 0.338 m

Gas Tanks Filled on a Daily Basis

𝐷 = (5,000,000𝐿

𝑦𝑟)(

1 𝑦𝑟

365 𝑑𝑦)(

1 𝑔𝑎𝑙

3.785 𝐿) = 3619 𝑔𝑎𝑙/𝑑𝑎𝑦

𝑉𝑡𝑎𝑛𝑘𝑠 = (3619 𝑔𝑎𝑙

𝑑𝑎𝑦)(

1 𝑐𝑎𝑟 𝑡𝑎𝑛𝑘

12 𝑔𝑎𝑙) = 302 𝑐𝑎𝑟 𝑡𝑎𝑛𝑘𝑠/𝑑𝑎𝑦

References

1 DiBiasio, D., and VP Process Development. Memo Report: African Centre for Renewable

Energy and Sustainable Technology. N.p., n.d. Web. 18 Nov. 2015. 2 Henley, Ernest J., Junior D. Seader, and D. Keith. Roper. "Chapter 7." Separation Process

Principles. Hoboken, NJ: Wiley, 2011. 384. Print.