The Effect of Material Choice on the Combustion Chamber of a Rocket Cooking Stove: Adobe, Common Brick, Vernacular Insulative Ceramic, and Guatemalan Floor Tile (Baldosa)

By Dean Still and Brad van Appel January 1, 2002

The Search for Vernacular Refractory Materials

Multiple tests of the Lorena stove beginning in 1983 at the Aprovecho Research Center have shown that placing thermal mass near the fire has a negative effect on the responsiveness and fuel efficiency of a cooking stove. In 1996, Leoni Mvungi built a Rocket stove from earth, sand, and clay that was a replica of a low mass Rocket consisting of metal chimney parts. His version weighed hundreds of pounds even though the Rocket internal chimney was only eleven inches high. Tests of a low mass sheet metal version scored around 30% fuel efficiency. But the best result achieved by the Mvungi stove was around 15%.

Building Rocket stoves from sand and clay showed little promise of improving on the three stone fire which was scoring around 18% in repeated boiling tests performed by Jim Kness and Dean Still (1994). Unfortunately, metal stove parts also have a major drawback in that the high heat in the combustion chamber quickly destroys thin metal. Consultants were in agreement that a good stove should last for years without requiring maintenance. Replacing metal parts as they wear out was not considered a viable solution.

A women's co-operative in Honduras (Nueva Esperansa) makes ceramic stove parts that have a reputation for working well in stoves. Aprovecho consultants Mike Hatfield and Peter Scott contracted with this group to produce combustion chambers for the Dona Justa plancha stoves that they helped to design. This material seemed to work well and, in fact, the Rocket elbow made by Nueva Esperansa has been successful in Honduras and Nicaragua. It is difficult, however, to deliver the fragile combustion chambers without breaking them. Also they are relatively expensive, costing about eight dollars each.

Don O'Neal (HELPS International) and Dr. Larry Winiarski have shown that cast iron combustion chambers, which do last, also have problems. Tests showed that the very conductive cast iron made the fire hard to start. In fact, a group of indigenous Guatemalan women stove testers living in Santa Avelina were unhappy with the expensive cast iron combustion chamber and asked for it to be replaced. They wanted a more responsive stove that started quickly, and quickly cooked food in the morning. Don and Larry eventually found an alternative material: an inexpensive Guatemalan ceramic floor tile (called a baldosa in Spanish) which seemed to function well when cut up to make the walls of the Rocket combustion chamber. The baldosa was about an inch thick so the combustion chamber only weighed eighteen and a half pounds. Like all Rocket combustion chambers it is surrounded by insulation, either wood ash or pumice rock.

The baldosa tile has done well in test stoves. It seems to be durable, lasting a year so far, and the group of testers from Santa Avelina reported that their stoves are much improved. The ceramic material made the stove much quicker to heat up. The women approved the improved stove for general dissemination to neighbors and other villages. The HELPS molded griddle stove now uses a preformed ceramic combustion chamber made by a local baldosa manufacturer. Unfortunately, all baldosa are not equally resistant to heat and it's important to test tiles before using them in stoves.

Appreciating that ceramic seemed a promising material for Rocket combustion chambers, Ken Goyer, an Aprovecho Board Member and consultant, spent a year, 2000-2001, testing ceramic mixes. His research resulted in a vernacular insulative ceramic material (VIC) that is refractory, insulative and can be home made. Six bricks made from this material combine to make a complete Rocket combustion chamber. Making the chamber from separate bricks has resulted in a greatly reduced tendency to crack. The bricks have held up so far in durability tests and they seem to create a very active fire.

The purpose of this paper is to describe the results of experiments involving same sized brick combustion chambers made from adobe, the insulative ceramic mix and common ceramic brick material. All bricks shared the same dimensions. Six bricks (11 ½" high by 2 ½" thick) made up a hexagonal cylinder surrounding a four inch in diameter chimney. Sticks of wood entered the bottom of the chimney through a hole sawn in the bricks. A combustion chamber made to similar dimensions was constructed using baldosa tile bought in Guatemala. Vermiculite filled in around the baldosa creating a combustion chamber with approximately the same dimensions as the brick stoves.

Protocols for Standard Stove Tests Using PICO Software

Twelve standard tests were conducted to test the response of four different combustion chamber materials: common brick, vernacular insulative ceramic (VIC), baldosa, and adobe.  Three tests were done on each material and an average performance was computed. Overall stove efficiency was also monitored during these tests. The results in this discussion should be treated as preliminary until more repetition substantiates any conclusions. Temperature sensors were placed as follows:

Sensor 1: Records ambient air temperature.  Sensor (Omega Type K Nextel Ceramic Thermocouple,-300 to 2300F.) suspended in air 3' from stove.
Sensor 2: Records water temperature.Sensor suspended in the center of a pot of water on the stove, 1/2" from bottom of pot. The cylindrical steel pot is nine inches in diameter and five inches high. A cylinder of sheet metal called a pot skirt surrounds the pot creating a gap of 3/16 inch through which hot flue gases pass.
Sensor 3:Records temperature inside stove body 1/2" from inner wall of fire chamber.  A hole is drilled from outside of the stove body 90 degrees to the right of the fuel magazine (as seen when facing the fuel magazine).  The hole is drilled at a level half the vertical distance from the top of fuel magazine to the top of the stove body and to a depth of 1/2" from the inner fire chamber wall.  Sensor is placed in hole and hole is filled with fire clay.
Sensor 4:Records temperature of stove body 1" from inner wall of fire chamber.  A hole is drilled from outside of the stove body 1/2" to the left of the hole for sensor 3.  The hole is drilled to a depth of 1" from the inner fire chamber wall.  Sensor is placed in hole and hole is then filled with fire clay.  In the case of the baldosa stove, where the baldosa is only one inch thick, the sensor is mounted on the outer face of the baldosa and within the insulation layer.
Sensor 5:Records temperature of stove body 2" from inner wall of fire chamber.  A hole is drilled from outside of stove body 1/2" to the right of the hole for sensor 3.  The hole is drilled to a depth of 2" from the inner fire chamber wall.  Sensor is placed in hole and hole is then filled with fire clay. In the case of the baldosa stove, the sensor is mounted outside the stove body and within the insulation layer 2" from the inner wall of the combustion chamber.
Sensor 6:Records temperature as flue gases exit through a 3/16th inch gap between the pot and a single layer sheet metal pot skirt surrounding the pot.  Sensor is suspended between (but not touching) skirt and pot 90 degrees to the left of the fuel magazine and 5" above the top of stove body.
Sensor 7:Samples fire temperature.Sensor is available for use to look for hot spots and randomly sample temperatures within the fire chamber. ( This data is not included in the averages used in this report.) Sensor 8:Sensor 8 serves as a spare sensor in the event that another sensor malfunctions.

Procedure: Weigh water and wood (5 lbs. of water (2265g), and 1.5 lbs. wood (680 grams).  Kiln dried Douglas Fir is cut into pieces 18" X 5/8" X 1/2" except for one 18" piece which is split finer to provide kindling for starting fire. Prepare fire and place pot on stove.  Place all probes while stove is at room temperature.  Begin recording.  One minute after recording has begun, light fire.  Once fire is going, feed sticks into fire five at a time, burning at the tips and pushed into fire as consumed. Continue recording until 7620 seconds after test is initiated.   Make note on file of: date; test number and description (i.e. 3rd Pico test, 2nd using adobe brick); weight of wood used; weight of water used; time from lighting fire until boiling; water remaining at end of test; time when last wood was fed into firebox, and weight of charcoal produced.  Also note any inconsistencies in procedure.

Protocols for Special Pico Stove Tests

To-date, the following additional tests have also been conducted:

Tests 6P20 & 6P21

Tests 6P20 and 6P21were designed to begin exploring temperature variations within the fire chambers of the adobe and insulative earth stoves.
Sensor 1: Records ambient air temperature.Sensor suspended in air 3' from stove.
Sensor 2: Records temperatures under fuel magazine shelf, 3" in from entrance to fire chamber.
Sensor 3: Records temperatures inside fuel magazine entrance. Hole is drilled through stove body to inner wall of entry to fuel magazine, 1" from bottom and 1" from entrance. Sensor is inserted all the way through the brick until the tip of the sensor is flush with the inner wall of the combustion chamber.
Sensor 4: Records temperatures at inner wall of stove 3" up from bottom of fire chamber. Hole is drilled through stove body 90 to the right of fuel magazine opening. Sensor is inserted all the way through the brick until the tip of the sensor is flush with the inner wall of the combustion chamber.
Sensor 5:Records temperatures at inner wall of stove 8" up from bottom of fire chamber. Hole is drilled through stove body 90 to the right of fuel magazine opening. Sensor is inserted all the way through the brick until the tip of the sensor is flush with the inner wall of the combustion chamber.
Sensor 6: Records temperatures at inner wall of stove 11" up from bottom of fire chamber. Hole is drilled through stove body 90 to the right of fuel magazine opening. Sensor is inserted all the way through the brick until the tip of the sensor is flush with the inner wall of the combustion chamber.
Sensor 7: Records exit temperatures at center of vertical chimney opening, level with top of stove.
Sensor 8: Samples fire temperature. Sensor is available for use to look for hot spots and randomly sample temperatures within the fire chamber

Test 6P22

An efficiency test was conducted on a low mass rocket stove made from a 5 gallon metal paint bucket with a sheet metal combustion chamber and insulated with tinfoil and temperature resistant fiberglass matting. Sensors checked temperatures under the fuel magazine shelf, at the chimney exit, and randomly within the fire chamber

Test 6P22a Bucket Stove

Sensor 1:Records ambient air temperature. Sensor suspended in air 3' from stove.
Sensor 2: Records temperatures under fuel magazine shelf, 3" in from entrance to fire chamber.
Sensor 3:Records exit temperatures at center of vertical chimney opening level with top of stove, under pot.
Sensor 4:Sensor suspended 1/2" from bottom in center of pot of water.
Sensor 5:Samples fire temperature.Sensor randomly samples temperatures within the fire chamber.

Results of Testing

The following four graphs show the average results of three tests using each of the four materials. They reflect how heat passed into the four materials as the one and a half pounds of wood was burnt. Temperatures rise much higher in the insulated materials (vernacular insulative ceramic and baldosa/vermiculite) but not as high in the denser types (common brick and adobe). Both the inner and outer sensors are hotter in the two more insulating samples. The sensors furthest from the fire in the common brick and adobe stoves reached peak temperatures of 320 F. and 222 F. compared to the hotter temperatures seen in the outer surfaces of the baldosa/vermiculite and VIC stoves, 497 F. and 343 F. At first glance this result seems counter intuitive. A perfect insulation wouldn't let any heat through the material.

Upon reflection, it makes sense that the better insulation would create higher temperatures near the inner faces of the chimney. Walls in a well insulated house should be close to room temperature because heat is passing so slowly though the wall that inner surfaces become almost as warm as interior air. Insulation is made up of pockets of air that cannot hold or pass a great quantity of heat. Denser materials 1.) require more Btu's to rise in temperature and 2.) conduct more heat through the material.

But, lighter materials, like air in insulation, respond much more rapidly to the effects of heat. Dense objects that are moderately conductive like common brick and adobe do not rise in temperature as fast as light weight, materials full of air holes. So, a reduced number of BTU’s of heat cause rapid and higher temperature rises in insulative materials. It is to be hoped that although the temperature rises faster and higher in an insulated brick many fewer BTU’s of energy are absorbed when compared to the heat absorption in a denser material. It would be reassuring to check to make sure that this is actually so:

1.) Average temperatures ½ inch from the fire within the vernacular insulative ceramic brick reached 906 degrees F. At the same place the baldosa/vermiculite brick rose up to 764 degrees F. but the sensors ½ inch within the adobe and common brick topped out at 427 F and 487 F. The more massive walls are much cooler.


In these tests the better insulator creates a graph that steeply rises to higher temperatures and the three lines are farther apart, i.e., heat passes more slowly through the material so there are bigger differences in the temperatures recorded at a increasing distance from the heat source. The maximum difference between the furthest apart sensors in the vernacular fire brick was 839 degrees F. In the baldosa/vermiculite test the maximum difference was 439 F. But in both the common brick and adobe combustion chambers the greatest difference was less than 300 degrees F. ( 275F. and 173F). The VIC brick seems superior in slowing down the passage of heat and it creates the hottest temperatures close to the fire.


2.) It is desirable that a combustion chamber stay hot for as long as possible. Retained heat in the combustion chamber helps when starting the fire and coals last longer in a hotter environment. At the end of the tests, after 127 minutes, the temperature closest to the fire one half inch  inside the vernacular insulative ceramic brick was 288 F. The common brick showed a temperature of 312 F. and the adobe was cooler at about 237 F. But the baldosa/vermiculite was hotter at 367 F. Like a good wall in a house the addition of some mass within the envelope of insulation seems to hold heat for the longest time. Even though the VIC brick got hotter and was a better insulator, in these tests the baldosa/vermiculite combination seems to perform better at retaining heat over time.


3.) Fuel efficiency was effected by mass but the influence of this amount of mass was not very powerful. The combustion chambers weighed in as follows:


·        Adobe                                                        50 ¼ pounds

·        Common Brick                                          36 pounds

·        Baldosa/Vermiculite                                  18 ½ pounds

·        Vernacular Insulative Ceramic                  16 pounds

·        Sheet Metal*                                                  3



The percentage of heat that entered the water in the pot follows a ranking by weight. But, although weights differed dramatically the difference in fuel efficiency, which was small, was obviously effected by other influences. The fuel efficiency of each material was as follows:


·        Adobe                                                  22.2%

·        Common Brick                                    22.4%

·        Baldosa/Vermiculite                            24.16%

·        VIC                                                      26.5%

·        Sheet metal low mass stove*               31.9%


*( Test 6P22a – Bucket Stove. A separate test included for the sake of comparison.)


The differences in fuel efficiency created by the four materials are not large. When using the boiling test all four ceramic materials do not perform badly. Each Rocket stove with any material in the combustion chamber did better than the laboratory tests of the three stone fire.( The pot skirt helps to raise efficiencies.) A larger difference is seen in an additional test of a sheet metal  combustion chamber which scored appreciably better than the higher mass types. But, so far, a long lasting, low cost, metal combustion chamber does not exist. Noting the success of the truly low mass sheet metal combustion chamber reinforces the design principle of lowering the mass around the fire.


4.) The responsiveness of the stove and the speed at which water boiled was effected by the material used. The five pounds of water boiled at the following times:


·        Adobe                                                       16.5 minutes

·        Common Brick                                         17.5 minutes

·        Baldosa                                                     19.2 minutes

·        Vernacular Insulative Ceramic                 12.7 minutes


The vernacular insulative ceramic made a much faster responding stove. If there is a demand for quick service then the first choice is obvious. Also when only making coffee or when boiling food that is rapidly prepared, the fuel savings could be significant. However, consultants report that most cooking in Central America is of a longer duration so this difference may not be reflected to a great degree in fuel saving. The attraction of the VIC brick is that stoves using this material should be quicker to light and faster to make hot fires.


5.) Does material choice effect combustion temperatures? A sensor was placed at the exit between the skirt and pot. But uncontrolled variations in the annulus gap seemed to distort these readings. No significant conclusions can be reached using this data.


For this reason, follow up tests were performed on the VIC, adobe and low mass metal combustion chambers. The stoves were fired as hot as possible without creating excess smoke or coals. During a 45 minute period temperatures were recorded using the PICO software at the following places in the adobe and VIC combustion chambers:

three inches, eight inches and eleven inches up from the bottom of the combustion chamber at the inside face of the Rocket elbow chimney and at the top of the chimney.



In the adobe combustion chamber the peak temperatures were as follows:


·        3"          1123 F.

·        8"            513 F.

·        11"          622 F.

·        Exit        1148 F.


The combustion chamber made from vernacular insulative ceramic brick showed the following peak temperatures:


·        3"         1383 F.

·        8"         1148 F.

·        11"       1113 F.

·        Exit      1573 F.


It was not feasible to drill into the low mass stove to replicate the tests done on the two ceramic stoves. But exit temperatures found during a similar test topped out at 1,592 degrees F.


Results from these tests indicate that an insulated combustion chamber is more likely to create high temperatures that should positively effect combustion efficiency. It is possible that higher combustion temperatures may decrease harmful emissions as well. If higher temperatures help to decrease emissions then the use of a more insulative material might recommend itself. The lower temperatures in the adobe would certainly seem to be detrimental to assisting secondary combustion within the Rocket chimney.



1.)    The internal surfaces close to the fire of both the vernacular insulative ceramic and baldosa/vermiculite get much hotter in less time than the brick and adobe. Hotter combustion chambers should be easier and more efficient to use.

2.)    The baldosa/vermiculite stays a bit hotter than the other materials which could have benefits for banking coals overnight. Also retaining heat makes the combustion chamber facilitate lighting a new fire, less heat is absorbed into the material, etc.

3.)    There isn’t a big difference in fuel efficiency as shown by the boiling test ( 22% to 26%). All materials do ok, but the lower mass models do tend to use less fuel.

4.)    Using the VIC brick results in a faster boiling time which should be appreciated by users. If the stove is used for short term cooking tasks then this material choice could result in significant fuel savings.

5.)    Material choice has a large effect  both inside the combustion chamber and on exit temperatures at the top of the Rocket elbow. The VIC combustion chamber showed maximum temperatures around 1575 F but the adobe model was quite cooler at 1148 F. Since higher temperatures assist more complete combustion then the use of a more insulative material should be beneficial.


Suggestions for Additional Research


1.)    Reducing the weight and increasing the insulative value of the VIC should increase fuel efficiency in Rocket stoves. But durability is very important in material choice for the combustion chamber. An improved VIC should also last for as many years as possible.

2.)    Baldosa/vermiculite seems to function well in Rocket stoves not because baldosa is very insulative but because the baldosa is relatively low mass and is backed up by a good insulator. The baldosa heats up quickly and makes the fire easier to start but the use of the VIC material seems to improve boiling rates. A thinner baldosa should perform better.

3.)    Using common brick slows down the response time of the stove and may make the stove more smoky but it does not seem to be significantly detrimental to fuel efficiency. If common brick is the only available option, it’s use seems to be justifiable.

4.)    Increasing combustion chamber temperatures can be achieved by using a VIC or baldosa/insulation material. Higher temperatures should result in a cleaner burn and in our tests these stoves did seem to produce less smoke. The combustion chamber bricks that were tested were 12 inches high. Studies at Aprovecho have shown that the most fuel efficient Rocket internal chimney is shorter, on the order of seven or eight inches high, because temperatures reaching the pot are hottest (above 1,600 F.). But this short chimney is smokier. A insulated internal chimney around 20 inches high is very clean burning but temperatures at the pot are greatly reduced (after 30 minutes around 700 F.) until equilibrium is reached. So chimney height is an important variable. The choice of 12 inches for these tests has created a relatively clean burning stove that is not as fuel efficient as possible in shorter term tests. Stove designers should take the decision of internal chimney height seriously.


Aprovecho Research Center
80574 Hazelton Road
Cottage Grove, Oregon 97424