bt analysis de proceso
DESCRIPTION
.TRANSCRIPT
Bioprocess Design and Economic Analysisfor the Commercial Production ofEnvironmentally Friendly BioinsecticidesFrom Bacillus thuringiensis HD-1 kurstaki
Gerald E. Rowe,* Argyrios Margaritis
Department of Chemical and Biochemical Engineering, Faculty of Engineering,The University of Western Ontario, London, Ontario, Canada N6A 5B9;telephone: (519) 661-2146; fax: (519) 661-3498; e-mail: [email protected]
Received 14 August 2003; accepted 27 October 2003
Published online 5 April 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bit.20146
Abstract: A production process for B. thuringiensis (Bt)bioinsecticides was designed in detail, including alternativebatch, low-density fed-batch (LDFB), and high-density fed-batch (HDFB) fermentation configurations. Capital andoperating costs, as well as profitability based on simplerate of return, were performed using a purpose-writtenFORTRAN program, explicitly analyzing production of awater-based flowable product used in forestry applications.
The total capital cost was $18 million (Canadian dollars)for a stand-alone plant with base-scale capacity of 3 � l07
billion international units (BIU)/year. Raw material costsamounted to $1.5 million yearly, of which approximatelyhalf was for formulation ingredients. Per-unit productioncost rose sharply for scales of less than 1 � 107 BIU/year,but was little affected by scale above 3 � 107 BIU/year.Product cost was much lower at all scales for a LDFB asopposed to batch fermentation process, but HDFB gaverelatively little additional cost benefit. Profitability analysisperformed by co-varying scale and selling price showed thatbreak-even occurred at a price of $0.45/BIU for a batchprocess at base scale, while with LDFB fermentation thesame production volume sold at $0.35/BIU gave a 12% rateof return. Since the assumed base scale would represent8–15% of current world Bt bioinsecticide production,based on value or volume, itwas concluded that profitabilitywould require some or all of the following elements: tar-geting higher-value markets such as disease vector control,in addition to forestry; a potentially lower plant capacity(although at least 1 � 107 BIU/year); and coproduction ofother large-volume microbial products to absorb capacityand match bioinsecticide output to market demand. B 2004Wiley Periodicals, Inc.
Keywords: bioinsecticides; bioprocess design; bioinsecti-cides production and costs; bacillus thuringiensis; eco-nomic visibility analysis
INTRODUCTION
Bioinsecticides derived from cultivation of Bacillus thu-
ringiensis (Bt) comprise over 90% of all biopesticides sold
on a world scale (Glare and O’Callaghan, 2000). These pro-
ducts provide effective, non-toxic, and environmentally
benign control of a wide range of lepidopteran, coleopteran,
and dipteran insect pests in agricultural, forestry, ornamental,
and disease-vector applications.
Based on a literature review, Glare and O’Callaghan
(2000) gave a 3-page list of commercial Bt products made in
dozens of countries, acknowledging that it was far from
complete. These materials derive from fermentation of
Bt subspecies kurstaki, aizawai, morrisoni (tenebrionis),
israelensis, and others that are formulated into a wide range
of solid and liquid product types. Some 26 Bt products are
registered with the United States Environmental Protection
Agency, and marketed worldwide as part of the US$l00
million annual global market for Bt bioinsecticides (Bishop,
2002). Annual world production of Bt bioinsecticides is
estimated to be approximately 13,000 tons (World Health
Organization, 2003a), with China and other developing
countries accounting for at least half of the total (Bishop,
2002). It is predicted that in the current decade the market
for biopesticides will grow by 10–15%, driven especially
by increased use of Bt for insect control in organic agri-
culture, forestry, and control of mosquito-borne diseases
(Bishop, 2002).
Despite (as well as because of) this commercial im-
portance of Bt bioinsecticides, little has been published on
their mass production and formulation (Couch, 2000). Even
less has been published on the economics of commercial
production, the only significant exception being the brief
generic analysis of product cost and gross profit for an
unspecified formulation type at scales of 20 and 200 tons
per annum done by Lisansky et al. (1993). Given this dearth
of information on industrial production of Bt bioinsecticides,
the significant scale on which they are used to control forest
B 2004 Wiley Periodicals, Inc.
Correspondence to: Argyrios Margaritis
*Present address: Biotechnology Research Institute, National Research
Council of Canada, 6100 Royalmount Avenue, Montreal, Quebec, Canada
H4P 2R2
Contract grant sponsors: The Natural Sciences and Engineering
Research Council of Canada; the EJLB Foundation
Contract grant number: OGP4388
378 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 86, NO. 4, MAY 20, 2004
insect pests in Canada (Van Frankenhuyzen, 2000), and as
a natural outgrowth of our years of research in this area
(Rowe andMargaritis, 1987, 1994, 2003; Rowe et al., 2003),
it was decided to perform a state-of-the-art bioprocess design
study and economic feasibility analysis for commercial
production of Bacillus thuringiensis bioinsecticides.
At the present time, there is worldwide interest in envi-
ronmentally friendly, non-toxic, and biodegradable bio-
insecticides produced by the soil bacterium B. thuringiensis
for use in agricultural crops (corn, soybean, wheat, cotton,
rice, vegetables, fruits, etc.), forestry, floriculture, and
disease vector control. Such interest arises fundamentally
from the serious, global-scale environmental pollution
problems caused by excessive use of chemical insecticides
during the last 50 years. Through this article, we intend to
assist in renewing interest in commercialization of B.
thuringiensis bioinsecticides, especially in developing
countries where native raw materials can be used to produce
effective biorational pest control agents.
MATERIALS AND METHODS
All prices stated are in Canadian dollars ($Cdn), taken as
equivalent to $US 0.73 for purposes of conversion.
General Process Description
Process analysis and economic evaluation of the production
of various formulations of B. thuringiensis bioinsecticides
was performed using a model created in FORTRAN. Both
batch and fed-batch fermentation methods were explicitly
modeled, while recovery and formulation operations for
various product types were kept invariant. A generic
flowsheet showing major unit operations for the process is
given in Figure 1. Although several items of equipment are
not shown, their cost was taken into account in the
economic analysis (see discussion below, and Table I).
With reference to Figure 1, soymeal is milled to
200 mesh in Substrate Hammer Mill C-110, and suspended
with corn steep liquor (CSL) in half the initial medium
volume in Fermentor R-400 before being sterilized with
steam. Corn syrup is diluted in Corn Syrup Mix Tank F-
210, then continuously preheated, sterilized by steam
injection and passage through Sterilization Holding Coil
R-320, and cooled before entering R-400. In fed-batch
configurations only, Medium Feed Tank F-460 is designed
to hold one half the volume of the initial medium in R-400.
Following fermentation, the stabilized broth is cooled to
10jC in Broth Cooler E-530 then adjusted to pH 4 with 5M
sulfuric acid in Broth Tank F-520. It is next concentrated
in disk-stack Centrifuge H-540, operating at a 20% of
its swallowing capacity and 2% solids loss (Villafana-
Rojas et al., 1996), to a paste containing 100 g biomass dry
matter/L. The paste is formulated in Primary Blending Tank
F-600, for production of either a primary powder inter-
mediate or water-based flowable final products (Lisansky
et al., 1993). Stream S-122 going to make primary powder
is dried in Spray Dryer D-620 and comminuted in Hammer
Mill C-621. D-620 was designed to reduce the moisture
content of the feed from 88.5% to 7.0%, assuming 35jCproduct outlet temperature and residence time of 42 s (Mas-
ters, 1991). Primary Powder Storage Bin F-630 is sized to
hold product from five fermentation runs. From there,
primary powder passes through Powder Weigh Hopper
F-632 (not shown), either to Wettable Powder Mixer X-650
for wettable powder production or to Secondary Blending
Tank F-640 for emulsification into an oil-based flowable
product. Fill and Weigh System X-610 is assumed to be
capable of packaging all three product types.
Raw Materials and Sterile Medium Preparation
Receiving and storage of medium ingredients, and medium
preparation and sterilization, comprise Sections 100, 200,
and 300 of the process (Fig. 1, Table I).
Medium composition was essentially that given by
Lisansky et al., (1993), comprising corn syrup as the main
carbon source, hammer-milled defatted soymeal and corn
steep liquor (CSL) as nitrogen sources, and 0.1% v/v
polypropylene glycol 2000 as antifoam. As indicated in
Figure 1, soymeal would be replaced with fishmeal in
B. thuringiensis subsp. israelensis fermentation for disease
vector control products (Lisansky et al., 1993); soymeal
cost was used for economic analysis. The fraction of
substrate dry matter from corn syrup, soy flour, and CSL in
the final medium for both batch and fed-batch config-
urations was assumed to be 0.526, 0.305, and 0.169,
respectively. Substrate concentrations were calculated as
follows: for the batch case, a total substrate dry matter
concentration of 33.1 g/L was assumed, corresponding to
the 18 g/L glucose concentration advocated by Lisansky
et al., (1993); for the two fed-batch cases total substrate dry
matter concentrations of 65.6 and 131.4 g/L were assumed,
in correspondence with the amount of glucose used in the
two most successful fed-batch runs reported by Liu et al.
(1994), designated by them as runs #6 and #7 (see below).
A detailed list of raw material usage and costs for the
base case is given in Table II. Delivered costs for raw
materials in appropriate delivery quantities were estimated,
assuming delivery cost of $30 per ton, using data either
directly from suppliers or the current chemical trade
literature (Chemical Market Reporter, 2003).
A combination of continuous and batch sterilization of
medium ingredients is employed (Lisansky et al., 1993).
Soy flour and CSL, in half the medium volume, are batch
sterilized for 4.5 h in Fermentor, R-400 (cf. Medium Feed
Tank F-460 in fed-batch configurations). After this has
Figure 1. Process flowsheet for production of a range of B. thuringiensis bioinsecticide formulations by batch or fed-batch fermentation.
ROWE AND MARGARITIS: COMMERCIAL PRODUCTION OF ENVIRONMENTALLY FRIENDLY BIOINSECTICIDES 379
Table I. Equipment list, capital cost (in thousands $Cdn), and power usage for base case production scenario (LDFB process; 3 � 107 BIU/year).
Description No. Capacity Installed cost $000
Section 100: Soymeal receiving, storage, and milling
F-100: Soymeal storage bin 1 30 cu m 12.3
X-101: Truck dump scale 1 91 tons 89.6
J-102: Receiving elevator 1 12 m 23.3
J-103: Auger 1 2.6 m 6.0
C-110: Substrate hammer mill 4 55 kW 32.2
J-111: Milling elevator 1 10 m 21.8
F-112: Surge hopper 1 3.5 cu m 6.2
H-115: Bag filter 1 2.0 cu m/s 83.0
F-116: Weigh hopper 1 33 cu m 25.3
Subtotal 299.7
Section 200: Liquid substrate storage and preparation
F-202: Corn steep storage tank 1 20 cu m 23.7
L-203: Corn steep pump 1 0.7 cu m/h; 0.4 kW 22.2
F-210: Corn syrup mix tank 2 105 cu m 90.5
M-211: Corn syrup mixer 1 8.4 kW 31.4
F-212: Corn syrup storage tank 1 61 cu m 42.0
L-213: Corn syrup pump 1 2.0 cu m/h; 1.1 kW 32.2
L-243: Medium sterilization pump 1 42 cu m/h; 16.5 kW 105.0
Subtotal 347.0
Section 300: Medium sterilization
E-300: Medium pre-heater 3 382 sq m 1,008.8
M-310: Steam injector 1 0.18 m ID 11.5
H-311: Steam carbon filter 4 4.6 sq m 50.1
H-312: Steam particulate filter 4 4.6 sq m 50.1
R-320: Sterilizer holding coil 1 0.31 m ID; 30 m 103.0
H-330: Flash cooler 1 0.88 mID; 1.6 m H 69.6
H-331: Entrainment separator 4 4.0 sq m 50.7
L-332: Hot medium pump 1 42 cu m/h; 16.7 kW 116.1
E-340: Medium cooler 1 33.4 sq m 160.3
Subtotal 1,620.2
Section 400: Fermentation
R-400: Fermentor 1 281 cu m; 322.6 kW 2,565.6
L-402: Broth pump 2 441 cu m/hr; 15.3 kW 127.8
R-410: Inoculum fermentor 1 14 cu m 380.8
F-420: Alkali storage tank 1 10 cu m 16.5
L-421: Alkali pump 1 3.1 cu m/h; 0.6 kW 27.9
F-460: Medium feed tank 1 94 cu m 65.1
M-461: Conc’d. medium mixer 3 19.4 kW 82.7
L-462: Conc’d. medium pump 2 7.5 cu m/h; 4.2 kW 90.1
Subtotal 3,356.5
Section 500: Product recovery
F-510: Sulfuric acid tank 1 61 cu m 67.4
L-511: Acid pump 2 5.0 cu m/h; 0.60 kW 26.5
F-520: Broth tank 3 95 cu m 1,267.4
M-521: Broth tank mixer 1 9.5 kW 65.3
E-530: Broth cooler 2 356 sq m 736.9
L-531: Broth pump 1 84 cu m/h; 5.9 kW 852.4
H-540: Centrifuge 3 420 cu m/h 2,322.9
L-541: Centrifuge feed pump 1 84 cu m/h; 5.9 kW 71.3
Subtotal 5,410.1
Section 600: Formulation and packaging
F-600: Primary blending tank 1 74 cu m 46.5
L-601: Primary blending pump 3 30 cu m/h; 5.9 kW 120.0
X-610: Fill-and-weigh system 1 200 L/min 172.3
D-620: Spray dryer 1 20 cu m 1,180.8
C-621: Powder hammer mill 1 54.7 Kw 15.8
F-630: Powder storage bin 1 41 cu m 14.5
J-631: Primary powder auger 1 3.0 m 6.0
F-632: Powder weigh hopper 1 7.0 cu m 18.1
F-640: Secondary blending tank 1 19 cu m 22.9
M-641: Emulsifier 2 12.4 kW 122.1
L-642: Blending tank pump 1 2.4 cu m/hr; 0.5 kW 23.8
(continued )
380 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 86, NO. 4, MAY 20, 2004
been cooled to 50jC, diluted corn syrup is sterilized in
high-temperature, short residence-time Sterilization Hold-
ing Coil R-320. In this operation the solution is heated to
90jC in Medium Pre-heater E-300 which also serves to
cool the sterilized liquid to 39jC. The preheat temperature
of 90 jC was chosen based on an approximate optimization
of heat exchanger vs. steam cost. Injection of clean, filtered
steam at 16 barg in Steam Injector E-320 heats the syrup to
140jC. After 3.0 min in R-320 the syrup is cooled to 98jCin Flash Cooler H-330, then cooled to 30jC by sequential
passage through E-300 and Medium Cooler E-340.
Fermentation Process Configurations
Fermentation operations per se comprise Section 400 of
the process (Fig. 1, Table I). Broth and product potency is
expressed in billions (109) of international units per unit
volume or mass viz. BIU/L or BIU/kg.
The fermentor, operating cycle of 66 h was assumed to
be similar to that of Lisansky et al. (1993), specifically
as follows: sterilization and checks: 1.0 h; medium load-
ing, sterilization, and cooling: 11.0 h; inoculation: 0.5 h;
fermentation: 38.0 h; product stabilization: 4.0 h; broth
discharge: 2.5 h. Inoculum volume was assumed to be l0%
of initial medium volume. Sparging with compressed air
was costed at a flow rate of 0.2 vvm, and mixer power
(electrical) assumed to be 2.1 W/L based on industry
practice at large scale. In fed-batch configurations the total
feed volume was assumed to be half the initial medium
volume. Culture conditions at harvest are typified by 90%
free spores and y- endotoxin crystals, with less than 10%
vegetative cells (Lisansky et al., 1993).
Fermentation biomass concentration and toxicity were
based on experimental data from Liu et al. (1994), which is
shown along with viable spore concentrations in Figure 2.
The biomass and spore concentration results of Arcas et al.
(1987) are also shown since these workers obtained the
highest reported Bt biomass concentration of 56.4 g/L, using
fed-batch fermentation. As the figure shows, the latter group
achieved biomass concentrations as high as 32.5 g/L in batch
fermentations, but with relatively low spore counts. The fed-
batch results of Liu et al. (1994) demonstrated that much
Table II. Raw material use and cost for base case production scenario (LDFB process; 3 � 107 BIU/year).
Raw materiala Use, ton/year Cost, $/ton Annual cost $000
Fermentation
Soymeal 560 320 178.0
Corn syrup 1,040 425 441.7
Corn steep liquor 560 100 56.1
Poly(propylene glycol) 30 190 5.8
Caustic soda, 50% 52 280 14.6
Subtotal 696.2
Product recovery
Sulfuric acid 300 85 25.8
Subtotal 25.8
Formulation
Poly(acrylic acid, Na salt), aqueous 180 1,580 285.4
Smectite clay 27 6,900 187.0
Xanthan gum 3.6 16,000 57.8
Non-ionic surfactants 30 3,500 105.4
Potassium sorbate 12 13,000 156.6
Subtotal 792.2
Total raw materials (thousands of $Cdn): $1,516.0
aExample vendors: soymeal: ADM Agri-Industries, Windsor, ON; corn syrup, corn steep liquor: Casco, London, ON; poly(propylene glycol), caustic
soda: Alphachem, Mississauga, ON; poly(acrylic acid): Polycryl, Stamford, CT; smectite clay: Elementis Specialties, Hightstown, NJ; xanthan gum:
NutraSweet Kelco, Chicago, IL; non-ionic surfactants, potassium sorbate: Ashland Canada, Mississauga, ON.
Description No. Capacity Installed cost $000
F-650: Oil storage tank 2 84 cu m 99.7
L-651: Formulation oil pump 1 4.2 cu m/h; 0.8 kW 28.6
Subtotal 1,871.1
Contingency and fee = $2,182.3
Capital cost of utilities, buildings, and site preparation = $3,637.1
Total capital cost: $17,943.2 Total continuous power: 1,060 kW
Table I. (continued )
ROWE AND MARGARITIS: COMMERCIAL PRODUCTION OF ENVIRONMENTALLY FRIENDLY BIOINSECTICIDES 381
higher spore concentrations are achievable for a given bio-
mass concentration. They also showed that nutrient feeding
could increase whole broth toxicity proportionally with
spore count at biomass concentrations up to 36.4g/L; at
a biomass concentration of 53.7g/L, however, broth toxi-
city was no higher and viable spore count fell precipitously.
These results also imply that the high-biomass fed-batch
results of Arcas et al. (1987) may have been suboptimal
in terms of broth toxicity. Based on these results the two
fermentations of Liu et al. (1994) giving the highest spore
concentrations (runs #6 and #7) were assumed to
be representative of results potentially achievable via fed-
batch fermentation.
For the batch fermentation case, it was assumed that
results using the equivalent of 18 g/L of glucose would be
directly proportional to those found by Liu et al. at 8 g/L
glucose, i.e., biomass concentration was assumed to be
13.3 g/L [(18/8) � 5.9 g/L], and broth toxicity to be
0.61 BIU/L [(18/8) � 2.72 � 108 IU/L]. The data of runs
#6 and #7 of these authors for the two fed-batch cases
exhibited biomass concentrations of 23.8 and 36.4 g/L, and
toxicities of 1.21 and 1.73 BIU/L respectively. These three
fermentation process configurations are designated hence-
forth as batch, low-density fed-batch (LDFB), and high-
density fed-batch (HDFB) processes, respectively.
Recovery and Formulation
Product recovery, formulation and packaging comprise
Sections 500 and 600 of the process (Fig. 1, Table I).
The plant was assumed to be capable of producing a range
of formulated final products for application in forestry,
agriculture, and disease vector control. It was designed to
produce three B. thuringiensis subsp. kurstaki formulations
for forestry and agricultural applications, namely: a water-
based flowable with activity of 12.4 BIU/L [i.e., the high-
potency ‘‘48B‘‘ product described by Lisansky et al.
(1993) for forestry applications], an oil-based flowable
with activity of 16.9 BIU/L, and a wettable powder with
activity of 16.0 BIU/kg. It was also designed to produce a
B. thuringiensis subsp. israelensis water-based flowable
formulation with activity equivalent to 8.45 BIU/L, used
for control of dipteran disease vectors. Detailed formulation
compositions were taken from Lisansky et al. (1993),
with the exception of the SurfynolR surfactants which were
judged to be prohibitively expensive at estimated bulk
delivered prices in the range of $16,000 to $18,000 per ton.
For economic analysis purposes it was assumed that a
combination of sorbitan monostearate and poly(oxyeth-
ylene) monostearate costing $7,000 per ton would give
satisfactory wetting performance. The formulation ingre-
dients used are shown in the list of raw material usage and
costs for the base case given in Table II.
Economic Analysis Calculation Methods
Economic analysis was performed in Canadian dollars for a
stand-alone plant located in southern Ontario, Canada, using
methods given in Ulrich (1984) unless noted. The plant was
assumed to operate 24 h/day, 330 days per year, resulting in
120 batches per annum with fermentation cycle time of 66 h.
The base case was taken to employ low-density fed-batch
fermentation yielding 1.21BIU/L, at a production scale of 3 x
107 BIU per annum, with all of the product formulated as a
water-based flowable for forestry applications and sold at an
average price of $0.35 per BIU. This price was chosen based
on reliable information that the current price of this type of
product to a large volume user is about $0.30 per BIU, and
$0.45 per BIU for small-volume purchases (Van Frank-
enhuyzen, 2003). Although a bioinsecticide plant would be
expected to produce all types of formulation for which a
market of at least a minimum volume exists, it was decided
to base the manufacturing cost and profitability analyses on
production of a water-based flowable only, due to a lack
of reliable price information for large-volume purchases of
other formulation types. The issue of production scale is
discussed below.
A FORTRAN program was written to calculate required
equipment size, number of identical units and installed cost,
total capital investment, operating costs, revenue, and over-
all profitability. The calculation procedures used are des-
cribed below, and an outline of the algorithms is given in the
Appendix. Program code was debugged, compiled and run
using WATFOR-77R Version 4.0 PC/DOS software
(WATCOM Systems Inc., Waterloo, Ontario).
Equipment costs were in general derived from Ulrich
(1984), and in some cases Peters and Timmerhaus (1980),
using a Chemical Engineering Plant Cost Index of 400.9.
Both sources give log–log graphs of purchase cost vs. unit
size from which the equations of linear segments of the
appropriate plots were estimated, generally over several size
Figure 2. Viable spore concentration results from [10] (5,n) and (o,.),and broth toxicity results from [10] (D,E). Open and closed symbols
represent batch and fed-batch fermentation results, respectively.
382 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 86, NO. 4, MAY 20, 2004
ranges, and coded to allow calculation of purchase cost as a
function of unit size. Installed cost for each piece of process
equipment (cf. Table I) was obtained by multiplying pur-
chase cost by an installation cost factor which varies with the
type of unit, construction material, etc. The installation cost
factor accounts for transportation of the equipment to the
site, direct materials (e.g., piping, valves, structural supports,
etc.) and labor, indirect expenses, contingency and fee
allowances, and capital for auxiliary facilities (e.g., steam
generation, etc.; Ulrich, 1984). All equipment in contact
with aqueous process liquids was costed on the basis of
316 stainless steel.
Two subroutines were used to perform these calculations
as the main program went through the equipment list,
namely: subroutine RANGE checked that the required size
was not greater than the maximum for the particular type of
equipment, and if it was, specified the number and size of
identical units; subroutine COST calculated the installed
bare module cost for the required size and number of units,
and accumulated these costs and electrical power require-
ments (continuous basis). The cumulative figure resulting
from this procedure was the total base bare module capital
required for the plant. To this value was added 18% for
contingency indirect costs for design and engineering,
construction expenses per se, contractual’s fee and con-
tingency (Peters and Timmerhaus, 1980). A further 30% of
total base bare module cost was added for the capital cost
of utilities, buildings and site preparation, to give an
estimate of total installed equipment cost. This type of
preliminary capital cost estimate, based on a detailed pro-
cess flow sheet and approximate material and energy
balances, is expected to be accurate to within 20% (Peters
and Timmerhaus, 1980; Ulrich, 1984). Total installed equip-
ment cost was multiplied by 1.09 to allow for start-up
costs, giving total fixed capital. Finally, working capital
(for raw materials and supplies, finished and semifinished
product, etc.) was added at 15% of total installed equipment
cost to give total capital investment.
Manufacturing cost and profitability calculations were
performed as described briefly below (cf. Table III,
Appendix). Raw materials usage and costs were summed;
packaging costs were assumed to be zero, i.e., product
supplied in bulk tanker or returnable bulk drums F.O.B.
plant. Operating labor consisted of four operators on each of
four shifts, each costed at $40,000 per annum; annual
administrative labor cost was assumed to be $120,000.
Individual costs were then calculated for steam generation
(natural gas assumed as energy source at a cost of $0.267 per
standard m3), electricity ($0.043 per kWh), cooling and
process water, demineralized water (for steam), refrigerated
water, and compressed air. Detailed trial calculations
showed costs for waste clarified broth disposal to be neg-
ligible. Total annual costs and profitability were calculated
as described in detail in the Appendix, assuming straight-line
capital depreciation over 15 years, zero initial research and
development costs, no tax concessions, and the current
Canadian federal tax rate of 21%.
RESULTS AND DISCUSSION
Installed Equipment and Total Capital Cost
Table I presents a listing of plant equipment, and total capi-
tal cost and power requirements, for base case production at
a scale of 3 � 107 BIU/year. For a stand-alone plant, total
capital cost was estimated to be almost $18million, of which
approximately $12 million is the total installed equipment
cost. The distribution of installed equipment cost. The
distribution of installed equipment cost among process
Sections 100 to 600 (see Table I) was found to be 2, 3,13, 26,
42, and 14%, respectively.Major items of equipment include
a fermentor of nearly 300 m3 volume, three large disk-stack
centrifuges, three broth-holding tanks, a spray dryer (not
required if all production were in fact water-based flowable),
and the medium preheater, accounting collectively for
approximately 69% of total installed equipment cost.
Production Cost Analysis
Usage and costs for raw materials for the base case are
shown in Table II. Raw material costs are estimated to total
approximately $1.5 million per annum of which the prin-
cipal carbon source, corn syrup, is the largest single cost
item (29% of the total). Nevertheless, even for a water-
based flowable product formulated using a low-priced sur-
factant, formulation ingredients comprise 52% of total raw
material costs.
Figure 3 shows unit production cost vs. annual production
rate for each of the three fermentation process configurations
considered, as well as a line corresponding to the assumed
selling price of $0.35 per BIU. It is seen that production cost
Table III. Manufacturing cost analysis for base case production scenario
(LDFB process; 3 � 107 BIU/year).
Expense item $000/year $/BIU
Raw materials 1,516 .051
Operating labor 640 .021
Administrative labor 120 .004
Utilities:
Steam: 16 barg @ $0.021 /kg
5 barg @ $0.016 /kg 881 .029
Electricity @ $.043 /kWh 398 .013
Process water @ $0.153 /cu m 592 .020
Demineraized water @ $.919 /cu m 101 .003
Cooling water @ $.007 /MJ 356 .012
Compressed air @ $.018 /std cu m 212 .007
Maintenance and repairs 978 .033
Operating supplies 147 .005
Laboratory charges 96 .003
Overhead and storage 521 .017
Local taxes and insurance 430 .014
Depreciation (15-year plant write-off) 1,304 .043
Administration 130 .004
Distribution and market development 349 .012
Total expense: $8,770 0.292
ROWE AND MARGARITIS: COMMERCIAL PRODUCTION OF ENVIRONMENTALLY FRIENDLY BIOINSECTICIDES 383
falls sharply, regardless of fermentation configuration, as,
production scale is increased in the range 5 � 106 BIU/year
to 1 � 107 BIU/year. Conversely for scales above about 3 �107 BIU/year, production cost does not decrease much as the
scale of operations is increased. It is also seen that increasing
fermentation broth potency from 0.61 BIU/L to 1.21 BIU/L
(i.e., going from batch-to low-density fed-batch operation)
has a strong effect in lowering unit production cost, while a
further increase to 1.73 BIU/L (i.e., HDFB operation) has
only a marginal effect on cost. For the two fed-batch cases
analyzed, production scale at the break-even point for
the assumed average selling price of $0.35/BIU occurs at
approximately 2 � 107 BIU per annum. Conversely, the
break-even point for the batch fermentation case, occurs at a
production scale of 6 � 107 BIU/year. This plot also shows
that for the fed-batch cases the actual selling price will have a
relatively strong effect on the scale at which the break-even
point is reached.
These results can be compared with total product cost
estimated by Lisansky et al. (1993), who found that 3.2 �106 BIU/year could be produced for $0.436/BIU [assuming
the ‘‘industry standard’’ (Lisansky et al., 1993) 16 BIU/kg
product and the usual currency conversion]. From Figure 3 it
is evident that, even if a HDFB fermentation process were
used, our estimate of total production cost is approximately
double their estimate for the same scale of operation. The
major difference between the two analyses appears to be
the magnitude of capital-dependent operating costs. Thus
Lisansky et al. (1993) project the same $/BIU capital-
dependent operating costs at a scale of 3.2� 106 BIU/year as
we estimate at a scale of 3 � 107 BIU/year. A judgment on
this discrepancy can be made by calculating the quantitative
variation of these costs with production scale found by the
two analyses, and comparing these to values reported in the
economic analysis literature. Accepted wisdom in this
regard is commonly known as the six-tenths-factor rule,
viz. ‘‘a log-log plot of capacity versus equipment cost for a
given type of equipment should be a straight line with a slope
equal to 0.6’’ (Peters and Timmerhaus, 1980). The capital-
dependent operating cost data for the two scales presented by
Lisansky et al., i.e., 3.2� 105 and 3.2� 106 BIU/year, give a
slope of 0.28 for log($/year) vs. log(BIU/year). For the
LDFB configuration of the present analysis with scale in the
range 0.5–6 � 107 BIU/year the slope for log($/year) vs.
log(BIU/year is 0.53 (r2 = 0.995), i.e., much closer to the
accepted value. It appears warranted to conclude that the
present capital cost estimates are considerably more realistic
than those of Lisansky et al. (1993), and hence to infer that
our production cost estimate is likely to be more realistic
than theirs.
The output from a detailed manufacturing cost analysis for
the base case is shown in Table III. Utilities as a group com-
prise the largest cost component, followed in decreasing
order by raw materials, depreciation, and maintenance and
repairs. Table IV shows operating costs by category as a
function of fermentation configuration at production scales
of 1 � 107 and 3 � 107 BIU/year. At both scales the fed-
batch processes gave significantly lower capital and utility
costs than the batch fermentation process, as well as some-
what lower administration and overhead costs, due to their
greater volumetric productivities. The lowest overall opera-
ting costs regardless of scale occurred for the high-density
fed-batch (HDFB) process, although raw material cost was
higher here than at somewhat lower productivity (LDFB).
As observed in both Figure 3 and Table IV, produc-
tion scale had a large effect on total per-unit operating
cost ($/BIU), which at 3 � 107 BIU/year was only 60–64%
of that at 1 � 107 BIU/year. Amounts in all cost categories
except for rawmaterials were substantially reduced at higher
scale, with the category showing the largest reduction
depending on the fermentation process configuration. Thus,
for batch fermentation by far the largest reduction was for
capital-dependent items, followed by labor-dependent items,
administration and overhead, and utilities. In the fed-batch
cases, the order for capital and labor dependent items
was reversed, reflecting the much lower capital cost of the
Table IV. Annual operating costs by category versus fermentation
configuration at two production scales (A: 1 � 107 BIU/yr; B: 3 �107 BIU/yr); all values in $/BIU.
Batch LDFB HDFB
Production scale A B A B A B
Direct fixed capital dependent items .203 .122 .139 .081 .122 .077
Labor dependent items .086 .028 .086 .028 .086 .028
Administration and overhead .120 .066 .094 .047 .087 .046
Raw materials .054 .054 .051 .051 .061 .061
Utilities .188 .144 .121 .083 .097 .064
Total .652 .415 .490 .292 .453 .279
Figure 3. Total per-unit production cost vs. production scale for batch
(solid line), low-density fed-batch (LDFB; dashed line), and high-density
fed-batch (HDFB; dotted line) fermentation processes.
384 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 86, NO. 4, MAY 20, 2004
latter type of process. For the LDFB fermentation process,
Figure 4 shows the effect of production scale on the per-
centage of total production cost for each of the cost cate-
gories in Table IV. It is seen, as expected, that as scale
increases raw material costs assume greater importance
while labor costs decrease as a percentage of total costs. It is
also seen that the cost of utilities becomes increasingly more
important as scale increases. Comparison of these results
with those given in Lisansky et al. (1993) for production of
3.2 � 106 BIU/year indicates agreement (at the same scale)
on utilities’ costs, but a much lower estimate in our case for
raw materials (about 7% vs. 21%) and a somewhat higher
estimate for capital costs, includingmaintenance (about 28%
vs. 23%). Although Lisansky et al. estimated utility costs at
$0.089/BIU vs. our value of $0.159/BIU, this factor accounts
for a similar percentage of total cost due to their low estimate
of the latter (see above). Their higher raw materials cost,
$0.090/BIU vs. our value of $0.05l/BIU, is no doubt due to
the higher priced formulation ingredients, especially surfac-
tants, which they specified (see the Materials and Methods
section). Finally, as discussed above, their $0.099/BIU ca-
pital cost estimate is unrealistically low compared to our
value of $0.213/BIU.
The present results can also be compared on a more
general basis with economic analyses of fermentation
processes found in the literature. One such analysis of a
comparable process at similar scale is that of Chong
and Lee (1997) for production of the biodegradable plas-
tic feedstock, poly(3-hydroxybutyrate). Their Alcaligenes
eutrophus process with product recovery by surfactant-
hypochlorite digestion at a scale of 24,800 m3 annual fer-
mentation broth volume was estimated to cost $US 640/m3
(about $880 in Canadian dollars). Our cost estimate to pro-
cess the same annual broth volume using LDFB fermen-
tation (3 � 107 BIU/year scale) is $3.54/m3. Given that the
former process requires much more elaborate operations
downstream of fermentation (i.e., three centrifugation stages
as well as spray drying), and that waste treatment/disposal
comprises 10% of their total costs, the production costs per
unit broth volume appear quite comparable.
Profitability Analysis
Profitability in terms of simple rate of return on total capital
invested was estimated for the three fermentation process
configurations by co-varying production scale and sale
price for a water-based flowable product. These parameters
were varied over the range 5 � 106 to 6 � 107 BIU/year,
and $0.18 to $0.50 per BIU respectively. For the batch
process, an average selling price of $0.45/BIU would be
required just to cover depreciation at the base scale of 3 �107 BIU/year. However the results for the fed-batch con-
figurations were much more promising.
The positive rate-of-return results obtained are plotted
for the LDFB process in Figure 5, and the HDFB process in
Figure 6. These plots show that at higher sale prices and
relatively low production scale the potency of fed-batch
fermentation broth (i.e., BIU/L) has only a slight effect on
profitability, while the effect becomes more substantial
as either price decreases or scale increases, or both. For
example, for LDFB the rate of return was 9.4% for 2 �107 BIU/year sold at $0.40/BIU and 9.5% for 4 �107 BIU/year sold at $0.30/BIU (Fig. 5), but 13.5 and
14.8% for the same combinations for HDFB (Fig. 6).
However, for 4 � 107 BIU/year sold at the assumed base
price of $0.35/BIU, the returns for LDFB and HDFB were
Figure 5. Rate of return on capital invested in a low-density fed-batch
(LDFB) fermentation process vs. product selling price and annual
production scale.
Figure 4. Percentage of total production cost due to various cost catego-
ries vs. production scale for a low-density fed-batch (LDFB) fermentation
process; (o) capital, (5) labor, (D) administration and overhead, (inverted
triangle) raw materials, (diamond) utilities.
ROWE AND MARGARITIS: COMMERCIAL PRODUCTION OF ENVIRONMENTALLY FRIENDLY BIOINSECTICIDES 385
15.3 and 22.1%, respectively. For the base case scale of 3 �107 BIU/year and sale price of $0.35/BIU, the profitability
of the LDFB and HDFB processes differs relatively little,
being 12.0 and 13.9%, respectively. Under the same
assumptions the rate of return for the batch fermentation
process is nil, viz. 0.1%. Thus, under the best-estimate sale
price and moderate production scale assumptions there
appears to be much to be gained by employing a low-
density fed-batch (LDFB) as opposed to a batch fermenta-
tion process. Conversely, there is little economic incentive
to push broth potencies and biomass concentrations to
values attainable by HDFB operation. Given the feed
control and aeration problems encountered by Liu et al.
(1994) at high biomass concentrations (cf. Fig. 2), process
robustness concerns further reinforce this conclusion.
Obviously it is necessary to have a realistic estimate of
howmuch of the various products could actually be sold (as a
function of price) to assess the prospects for commercially
viable production of Bt bioinsecticides. The Canadian Bt
market is overwhelming for control of forest insect pests:
during the 5-year period 1998–2002, annual Bt usage ranged
from approximately 6 to 12.5 � 106 BIU on an area of
208,000 to 353,000 ha, with average usage of some 8 �106 year (Van Frankenhuyzen, 2003). During the past few
years operational use of Bt in Canadian forestry has
decreased considerably because the spruce budworm out-
break that accounted for most spraying since the mid-1980s
has collapsed throughout eastern North America. However,
from 1985 to 2000 an average of approximately 1� 107 BIU
were applied annually, and during the worst years of the
spruce budworm infestation, from 1985–1991, yearly usage
was 1.4� 107 BIU (van Frankenhuyzen, 2003). Assuming a
return to these historic usage rates in the future, Canadian
forestry demand could amount to almost half of the 3 �107 BIU/year taken as base case output above.
From the perspective of total world Bt bioinsecticide
production of about 13,000 tons (World Health Organ-
ization, 2003a), assuming the industry-standard potency of
16 BIU/kg, base scale output would be 15% of present
global production. Given that about half of world Bt pro-
duction occurs in developing countries such as China
(Bishop, 2002), precisely to avoid having to import rela-
tively expensive inputs such as pesticides from the indus-
trialized countries, it seems more relevant to consider that
base case output would amount to some 30% of the current
market in the latter countries. It would appear unrealistic to
expect that such a large increase in output could be
profitably absorbed, whether or not anticipated modest
growth in market size materializes (see below).
In terms of value, global sales of Bt bioinsecticides are
estimated at US$ 100 million (Bishop, 2002), or in
Canadian dollars approximately $137 million. For the base
case scenario of 3 � 107 BIU sold at an average price of
$0.35/BIU, total revenue would be $10.5 million, which is
approximately 7.7% of the present value of world sales.
Given that this market is expected to expand at a rate of
1.0–1.5% per annum in the foreseeable future (Bishop,
2002), market growth could absorb these increased sales
within a period of 5–8 years, assuming that no additional
competing products were brought to market during that
time. Although this simple calculation ignores questions of
which market sectors will actually grow at such a rate, in
terms of geographic location and product type (target
insects and formulations), it does indicate that sufficient
global market demand could exist not long after the time
that a new plant was actually in production.
Although no figures are publicly available, one Bt
market sector that has been growing in recent years is that
of mosquito control to prevent the spread of diseases such
as West Nile virus, dengue fever, and malaria (Anonymous,
2003); one authority estimates it to comprise at least 20%
of global Bt use (Van Frankenhuyzen, 2003). It is also
appears to be a sector in which prices are higher than, for
example, the forestry market. One apparently large-volume
purchase of AquabacR B. thuringienis subsp. israelensis
granules, having a potency equivalent to 0.2 BIU/kg, was
priced at $US64.00 per 40 lb bag (Commonwealth of
Massachusetts, 2003). This price, about $24/BIU in
Canadian funds, is 80 times that of the $0.30/BIU price
which prevails in the bulk forestry market, and is probably
representative of prices in other ‘‘specialty’’ market sectors
such as horticulture, etc. However, it is likely that a
substantial fraction of Bt disease vector control products
have historically been sold in very large-volume purchases
at correspondingly lower prices to the World Health
Organization for its 1974–2002 Onchocerciasis Control
Programme (World Health Organization, 2003b). In sum-
mary, it is not possible to make a quantitative price esti-
mate for Bt products in the disease vector control market,
although it is certainly higher than that of the forestry
Figure 6. Rate of return on capital invested in a high-density fed-batch
(HDFB) fermentation process vs. product selling price and annual
production scale.
386 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 86, NO. 4, MAY 20, 2004
market (Van Frankenhuyzen, 2003). Thus, in conclusion,
no more realistic profitability estimates than those pre-
sented in Figures 5 and 6 are possible without a thorough
study of the entire range of Bt market sectors, which is well
beyond the scope of the present work.
It is possible, however, to conclude from the above
that profitable production of Bt bioinsecticides in Canada
will require a strategy involving some or all of targeting
higher-price markets, reducing production scale below the
assumed base case, or coproduction of other microbial
products in a shared facility. The figures cited above for
world Bt market value and production quantity can be
evaluated to give an average selling price of almost $0.70/
BIU, implying that much of this output is sold in con-
siderably higher value markets than that of Canadian
forestry (viz. bulk price of $0.30/BIU). From Figure 3 it is
evident that, if this average world price were attainable, by
selling a substantial fraction of the output into higher value
agricultural, ornamental, and disease-vector markets, Bt
production at a scale of somewhat less than 1 � 107 BIU/
year could be quite profitable, provided that at least a low-
density fed-batch fermentation process were employed.
Another, possibly complementary, alternative would be to
build a plant with a capacity of at least 1–2 � 107 BIU/year
to maintain economies of scale, but to produce other
microbial products as well. One such co-product might be
Rhizobium or Azospirillium inoculants that appear to
have substantial potential for increasing the efficiency of
utilization of fertilizers and the yields of both leguminous
and nonleguminous crops (Okon and Labandera-Gonzalez,
1994; Rao, 2003; Thomas et al., 1995). Another alternative
could be the biodegradable plastic, poly(3-hydroxybuty-
rate), which appears to have a scale similar to B. thu-
ringiensis bioinsecticides for economical production (Chong
and Lee, 1997).
CONCLUDING REMARKS
With reasonable care it should be possible to generalize the
present results to any fermentation-based process in which
the amount of product formed is proportional to biomass
concentration. For the three configurations analyzed above,
this quantitative relationship is (g biomass dry matter per
liter) = (20–22) � (BIU/L). For the LDFB configuration
producing 24,800 m3 of fermentation broth annually,
production cost was estimated at about $350/m3, compared
to $880/m3 for a biopolymer production process comprising
several more downstream processing operations. This
difference underlines the relative importance of post-
fermentation processing costs, especially primary recovery
operations, which accounted for 42% of our total operating
costs (cf. their waste disposal costs). The breakdown of our
base case operating costs in terms of major process
operations was calculated to be approximately as follows:
medium preparation, 20%; fermentation, 25%; primary
recovery, 40%; finishing, 15%. The detailed process design
and results presented herein, especially if critically
compared to the economic analysis of biopolymer produc-
tion by Chong and Lee (1997), should make it possible to
quickly perform an estimate of about F 30% accuracy of
the production cost of an arbitrary, semi-purified, biomass-
associated, submerged fermentation product.
The capital cost estimation approach implemented in the
present study, based on a detailed technical analysis and
design of the Bt bioinsecticide production process, appears
to be capable of generating capital cost estimates of suf-
ficient accuracy (i.e., F20%) for preliminary budget
approval purposes (Peters and Timmerhaus, 1980). The
methodology developed to obtain such estimates for var-
ious operating configurations and production scales con-
sists of a purpose-written computer program employing
piecewise log–log correlations of purchase cost vs. unit
size, combined with unit-specific installation cost factors
(Ulrich, 1984). The program also calculates requirements
for raw materials and utilities via static simulation of each
particular process. This approach, especially if the detailed
process can be faithfully modeled using a commercially
available process design software package, is a cost-
effective way of obtaining capital and operating cost
estimates for preliminary budget approval purposes. Of
course, the reliability of the results is crucially dependent
on the accuracy of the major assumptions, such as the
volumetric productivity of fermentation. To proceed with
any measure of confidence to estimate the profitability of a
facility it is necessary to have a reliable estimate of average
selling price, which for Bt bioinsecticides means price as a
function of the quantities sold in typical transactions, for
each of the various product market sectors.
APPENDIX
Outline of Economic Analysis Algorithm
Input process and economic parameters annual production,
BIU/year, fermentation configuration (batch or fed-batch);
broth potency, BIU/L; total medium substrate concentra-
tion, g dry matter/L; broth biomass concentration at har-
vest, g dry matter/L; selling price of each final product, $/L
or $/kg.
Calculate required fermentor broth volume per batch:
based on annual production, number of batches per year
(120), broth potency and product loss during recovery (2%).
Go through the equipment list sequentially for the
specified process configuration, performing the following
for each unit:
� Calculate size and required electrical power (continuous
basis)� Via subroutine RANGE, check that size does not exceed
maximum for type of unit; if it does, calculate the num-
ber and size of identical units� Via subroutine COST, calculate installed bare module
cost, and accumulate these costs and electrical power
requirements.
ROWE AND MARGARITIS: COMMERCIAL PRODUCTION OF ENVIRONMENTALLY FRIENDLY BIOINSECTICIDES 387
Multiply the total base bare module cost by 1.48 to
account for contingencies, fees, and the capital cost of
utilities, buildings and site preparation, to give an estimate
of total installed equipment cost.
Add 9% to total installed equipment cost for start-up
costs, giving total fixed capital investment.
Add 15% of total installed equipment cost for working
capital, giving total capital investment.
Calculate and sum raw material costs.
Calculate operating and administrative labor costs.
Calculate costs for each of low-and high-pressure steam,
electricity, water (cooling and process, demineralized refri-
gerated), and compressed air.
Calculate total direct costs by summing the costs for raw
materials, labor, utilities, maintenance and repairs (5% of
total fixed capital), operating supplies (15% of maintenance
and repairs), and laboratory costs (15% of operating labor).
Calculate overhead and storage costs as 30% of the total
of labor, and maintenance and repairs.
Calculate local taxes and insurance costs as 2.2% of total
fixed capital.
Calculate total indirect costs as the sum of overhead and
storage costs, plus local taxes and insurance costs.
Calculate total manufacturing cost, excluding deprecia-
tion, as the sum of total direct and total indirect costs.
Calculate annual depreciation as fixed capital investment
divided by 15.
Calculate administrative costs as 25% of overhead and
storage costs.
Calculate distribution and market development costs as
5% ot total manufacturing cost.
Sum administrative, and distribution and market devel-
opment costs, to get total general expenses.
Sum total manufacturing cost, depreciation, and total
general expenses to get total expense (total annual costs).
Calculate annual revenue assuming all production sold at
specified price.
Subtract total annual costs from annual revenue to get net
annual profit before taxes.
Calculate income taxes as 21% of net annual profit, and
subtract from the latter to give net annual profit after taxes.
Calculate aftertax rate of return, in %, as 100 times the
sum of net annual profit after taxes plus depreciation,
divided by total capital investment.
The authors wish to thank Dr. K. Van Frankenhuyzen of the
Canadian Forest Service, Sault Ste. Marie, Ontario for valuable
insights into the Bt forestry market in Canada; and S. Dodds for
implementing the process diagram in BioProR Designer. Dr. A.
Margaritis thanks NSERC; EJLB for their financial support.
References
Anonymous. 2003. Valent Biosciences Corporation acquires Certis
USA’s public health business; company strengthens leadership
position in Bt insecticides. PR Newswire, March 12, 2003, http://
www.findarticles.com/cf_0/m4PRN/2003_March_12/98638052/arti-
cle.jtml, October 1, 2003.
Arcas J, Yantorno O, Ertola R. 1987. Effect of high concentration of
nutrients on Bacillus thuringiensis cultures. Biotechnol Lett 9:
105–110.
Bishop A. 2002. Bacillus thuringiensis insecticides. In: Berkeley R,
Heyndrickx M, Logan N, De Vos P. (eds.). Applications and syste-
matics of Bacillus and relatives. Oxford: Blackwell. pp. 160–175.
Chemical Market Reporter. 2003, Feb. 24. New York: Schnell Publishing.
Chong J-I, Lee SY. 1997. Process analysis and economic evaluation for
poly(3-hydroxybutyrate) production by fermentation. Bioprocess Eng
17:335–342.
Commonwealth of Massachusetts. 2003. http://www.mass.gov./osd/memo/
updt0108.pdf, March 18, 2003.
Couch TL. 2000. Industrial fermentation and formulation of entomopa-
thogenic bacteria. In: Charles J-F, Delecluse A, Nielsen-LeRoux C.
(eds.). Entomopathogenic bacteria: from laboratory to field applica-
tion. Dordrecht: Kluwer. pp. 297–316.
Glare TR, O’Callaghan M. 2000. Bacillus thuringiensis: Biology, ecology
and safety. Chichester: Wiley. 350 p.
Lisansky SG, Quinlan R, Tassoni G. 1993. The Bacillus thuringiensis
production handbook. Newbury: CPL Press. 125 p. U.K.
Liu W-M, Bajpai R, Bihari V. 1994. High-density cultivation of
sporeformers. Ann NY Acad Sci 721:310–325.
Masters K. 1991. Spray drying handbook, 5th ed. Harlow: Longman. 725 p.
Okon Y, Labandera-Gonzalez CA. 1994. Agronomic applications of
Azospirillium: An evaluation of 20 years worldwide field inoculation.
Soil Biol Biochem 26:1591–1601.
Peters MS, Timmerhaus KD. 1980. Plant design and economics for
chemical engineers. 3rd ed. New York: McGraw-Hill. 973 p.
Rao DLN. 2003. Challenge program on biological nitrogen fixation. http://
www.icrisat.org/bnf/Abs2.htm, July 24, 2003.
Rowe GE, Margaritis A. 1987. Bioprocess developments in the production
of bioinsecticides by Bacillus thuringiensis. CRC Crit Rev Biotechnol
6:87–127.
Rowe GE, Margaritis A. 1994. Endocellular fatty acid composition during
batch growth and sporulation of Bacillus thuringiensis kurstaki. J Ferm
Bioeng 77:503–507.
Rowe GE, Margaritis A. 2004. Enzyme kinetic properties of a-1, 4-
glucosidase in Bacillus thuringiensis. Biochem Eng J 17:121–128.
Rowe GE, Margaritis A, Wei N. 2003. Specific oxygen uptake rate
variations during batch fermentation of Bacillus thuringiensis
subspecies kurstaki HD-1. Biotechnol Progr 19:1439–1443.
Thomas J, Palaniappan S, Hopper W, Nirmala CB. 1995. Development of
Rhizobium and Azospirillium inoculants with enhanced potential for
field application. In: Tikhonovich IA, Provorov NA, Ramonov VI,
Newton WE. (eds.). Dordrecht: Kluwer, Netherlands. pp. 659–664.
Ulrich GD. 1984. A guide to chemical engineering process design and
economics. New York: Wiley. 472 p.
Van Frankenhuyzen K. 2000. Application of Bacillus thuringiensis in
forestry. In: Charles J-F, Delecluse A, Nielsen-LeRoux C. (eds.).
Entomopathogenic bacteria: From laboratory to field application.
Dordrecht: Kluwer. pp. 371–382.
Van Frankenhuyzen K. 2003. Canadian Forest Service, Sault Ste. Marie,
Ontario, personal communication.
Villafana Rojas J, Gutierrez E, De la Torre M. 1996. Primary separation of
the entomopathogenic products of Bacillus thuringiensis. Biotechnol
Progr 12:564–566.
World Health Organization. 2003a. Environmental Health Criteria 217,
Bacillus thuringiensis. http://www.inchem.org/documents/ehc/ehc/
ehc217.htm, March 11, 2003.
World Health Organization. 2003b. Onchocerciasis (River blindness), Fact
Sheet No. 95, Revised February 2000. http://www.who.int/inf-fs/en/
fact095.html, July 24, 2003.
388 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 86, NO. 4, MAY 20, 2004