Few areas of fish processing generate more discussion and contradictory advice than the smoking process. Even though people have been smoking fish for centuries, the smoking process still remains more of an art than a science. The purpose of this paper is to cut through some of the fog and establish what we know for sure about the cooking process. In addition, I will present some helpful techniques that you can use in your fish smoking operations.

What is Hot Smoking? The terms cold smoking and hot smoking mean different things to different people. Scientists usually define cold smoking as a cooking process where the dry bulb temperature is kept lower than 86°F (30°C) for the entire process (Horner, 1992; Regenstein and Regenstein, 1991). Practically speaking, however, hot smoking is simply a heating process that fully cooks the fish to 140°F core or higher, while the cold smoking process does not (Hilderbrand, 1993). Fish that are prepared using either the cold or hot smoking processes have a distinctly different flavor, texture and appearance (Whelan, 1982). Cold smoking usually requires days or even weeks to complete the process, while hot smoking usually requires only a few hours.

Oven Design Batch processing ovens are an integral part of any smoked fish operation. Hot smoking is typically carried out in either a free- or forced-convection batch oven. The term 'forced-convection' simply means that air is circulated through the oven using a fan. In a free-convection oven, the air is not forced to circulate and moves only by natural movement of the hot air (Hallstr`m et al., 1988). Both free- and forced-convection ovens are often referred to as smokehouses.

In addition to the oven cabinet itself, most modern forced-convection batch ovens are equipped with certain basic components including a control system, heat source (electric, gas, or steam), humidity source (steam or water), main fan, exhaust fan, fresh air and exhaust dampers, rotating dampers, and temperature sensors. In addition to these components, auxiliary pieces of equipment such as smoke generators, liquid smoke systems, and cleaning systems are often included.

Batch ovens come in many different shapes and sizes, but regardless of their size, configuration or age, the ovens are designed to control four basic variables; cooking time, dry bulb temperature, wet bulb temperature, and air flow. Accurate measurement and control of these four variables is essential for the development of effective cooking processes. It is the control system's job to measure and control these critical variables, and to do so, it must be properly calibrated and in good working order.

Dry Bulb Temperature The dry bulb temperature is simply the temperature of the air as measured by a clean, dry temperature sensor (Fellows, 1988). The dry bulb sensor is usually installed in the top center of the oven. In this location, the dry bulb sensors measures the temperature of the return air that has already passed through the product and is being returned to the fan plenum to be reheated and re-humidified. In some oven designs, the dry bulb sensor is attached to one of the side walls, where it measures the temperature of the supply air that has not yet passed over the product. The control system regulates the dry bulb temperature by controlling the heat source, which may be gas, steam, or electric heat.

Wet Bulb Temperature The wet bulb temperature in an oven is measured by fitting a wet, moisture-wicking cloth over a temperature sensor and placing it in the oven air stream. The wet bulb cloth is draped in a pan of water to keep it moist, and is located next to the dry bulb sensor. When dry air is blown over the sensor, water evaporates from the cloth and cools the sensor (Watson and Harper, 1988). If the oven air is saturated (100% relative humidity), no evaporative cooling will take place, and the dry and wet bulb temperatures will be the same (Horner, 1992). The effect of evaporative cooling is readily observed by wetting your finger and blowing on it. The difference between the dry and wet bulb temperatures can be used to calculate the relative humidity.

The wet bulb temperature is essentially a measurement of the temperature at which moisture is evaporating in the oven. Since the surface of the fish is wet or moist, the wet bulb temperature has a strong influence over the product surface temperature and heating rate for much of the cooking process (Skj`ldebrand, 1980).

The wet bulb temperature in an oven can be controlled in two ways. The control system can either modulate fresh air and exhaust dampers open and closed to regulate the amount of evaporated moisture from the product that is retained in the oven. Or it can control the wet bulb temperature by injecting steam or atomized water into the oven as a humidity source.

If a processing application requires the oven to maintain relatively low temperatures, say 110°F dry bulb or below, steam humidity may add too much heat to the oven, causing the dry bulb temperature to creep above the setpoint. If so, the use of atomized water instead of steam for humidity will help to maintain a low dry bulb temperature setpoint.

Airflow Patterns A typical batch oven design and air flow pattern is shown in Figure 1.

The main fan and the heat and humidity sources are typically either located on top of the oven (as shown) or at the back of the oven. The main fan supplies the air to the supply ducts, where it is delivered to the process cabinet through supply cones or slots. The air travels down the side walls, along the floor, up through the product to the return duct, and then back to the fan cabinet to be reheated and re-humidified.

The location in the oven cabinet where the two opposing air streams meet is known as the 'break point'. In Figure 1, the break point is shown at the bottom center of the oven. As shown in Figure 2, many batch ovens include a rotating damper system that slowly sweeps the break point from side to side in the oven.

The rotating damper system consists of a set of dampers located in the supply legs that, while the oven is running, slowly rotate at approximately one revolution per minute. The sweeping action of the air from side to side in the oven serves to reduce the side to side temperature and shrink variation for the product.

The Cooking Process Hot smoking of fish in a forced air convection oven is essentially a high temperature drying operation (Skj`ldebrand, 1980). Fish cooking processes are usually multiple step cooking processes that may include drying, smoking and cooking steps. An example of a typical hot smoking process for salmon fillets is listed in Table 1.

Power of Evaporative Cooling A load of skin on salmon fillets (1 3/4 lb/fillet) was cooked and smoked in a forced air convection oven using the process listed in Table 1. Heating curves for oven dry and wet bulb temperatures and fillet surface and core temperatures are shown in Figure 3.

The surface temperature was measured by inserting a 1/16 inch diameter temperature probe just underneath the flesh side of the fillet. The probe was positioned such that the sensing tip of the probe was visible just underneath the surface.

As shown in Figure 3, the dry and wet bulb temperatures, and the process step changes were controlled according to the setpoints listed in Table 1. The fillets were dried for two hours at 110° in Step 1. For Steps 2 and 3, the dampers were closed and the product was smoked for two hours using traditional smoke.

The dry bulb temperature was increased from 110° in Step 2 to 150° in Step 3 to help develop and set the smoked color. After smoking, the dry and wet bulb temperatures were increased in Step 4 to finish cooking the product.

Throughout the cooking process, the surface and core temperatures were only approximately 6° apart (Figure 3). The reason that the surface and core temperatures were so close together is that a salmon fillet is fairly thin, and so it takes only a short time (approximately 10 minutes) for the heat to transfer from the surface to the core. However, even though a fillet could easily be heated to a fully-cooked core temperature in less than an hour, it still takes 3 to 5 hours to cook and smoke salmon fillets. In addition to cooking the product to the target core temperature, processing time is needed to condition the surface, allow it to absorb smoke and develop color and firmness.

In Figure 3, you can see that at the beginning of the process, the surface temperature rose very rapidly. However, after it moved through the wet bulb temperature (20 to 30 minute), the rate of heating slowed dramatically. This reduced heating rate was caused by evaporative cooling (Godsalve et at., 1977; Skj`ldebrand, 1980).

During the early stages of the process, the surface was very wet. Evaporation of the surface moisture cooled the product surface, resulting in a slower rate of heating. During the later stages of the process, as the surface gradually dried, the evaporative cooling had less effect on the product heating rate (Figure 3).

Since fish naturally contain water and have usually been soaked in brine, evaporation of moisture from the product surface makes the surface act much like a wet bulb sock (Hanson, 1988). For this reason, the wet bulb temperature has a strong influence on the product temperature and heating rate throughout the process - especially during the early stages of the process when the surface is very wet.

Importance of Drying In a typical hot smoking process of either whole fish or pieces, the product is usually dried at a low temperature (100 to 120° ) before being smoked and cooked. For example, the process shown in Table 1 has a two hour drying step at 110° in Step 1 before the smoke in turned on in Step 2. The drying process serves several purposes.

The primary purpose of the drying step is to dry some of the moisture off of the product surface so that the fish absorbs the right amount of smoke to create the smoked color that you want. The process in Table 1, for example, was designed to create lightly smoked salmon fillets, and so the product was dried at a low temperature for a fairly long time (2 hours) before smoke was applied. Since smoke is absorbed into the moisture on the surface of the fish, the more it is dried, the less smoke will be absorbed. If the fish surface is not dried at all before smoking, the wet surface will absorb a lot of smoke and will develop a dark brown smoked color. On the other hand, if the fish is over dried, it will not be able to readily absorb smoke, resulting in a smoked color that may be too light. By varying the drying time of the fish before smoking, you can very easily control the smoked color of the fish. For example, the color of salmon fillets can vary from a dark brown (or even black) smoked color if smoked with a wet surface, to a deep red smoked color if smoked with a moist (tacky) surface, to a light golden smoked color if smoked with a dry surface. The amount of surface moisture has a strong effect on smoked color, and is further discussed in the next section (The Smoking Process).

If you want to see the what effect various drying times will have on the smoked color of your own product, it is very easy to test during regular production runs. No complicated test procedures are required. When you run a production process, simply remove a few pieces from the load to serve as test pieces. Load the oven (less the test pieces) and start the process. After the drying period for the full load has run for a set period of time (for example, one hour of the two hour drying period for the Table 1 process), open the oven door and put in the test pieces. This way, the test pieces will have had less drying time than the production fish, and you can see the impact of the additional moisture on the smoked color. After the process is completed, you can compare the test and production product side by side to see the effect of the shorter drying period on smoked color. This type of test can be used to evaluate new process changes, or simply for training purposes.

Another purpose of the drying step is to firm up the fish before subjecting it to higher temperature cooking steps. The initial drying step is especially important to firm up whole fish what are hung from sticks during cooking, and helps prevent splitting and cracking of fillets. Whole fish that are inadequately dried before cooking will often fall apart later in the cooking process (Regenstein and Regenstein, 1991). Drying also helps form the glossy surface appearance often called the pellicle. The glossy appearance of the pellicle is considered a desirable quality for many fish products. And finally, a low temperature drying step helps to prevent formation of the white curd that sometimes oozes out of the fish during cooking (Regenstein and Regenstein, 1991; Whelan, 1982). This curd is especially noticeable on products like salmon fillets, where the white curd shows up dramatically against the pink background. Not only is the white protein curd unsightly, but it is also an excellent medium for bacterial spoilage.

Smoking Processes Traditional smoke is usually produced by smoldering sawdust in a hot plate smoke generator. Liquid smoke is usually purchased commercially, and is produced using a water extraction process of wood smoke vapors.

Smoke Composition Traditional wood smoke consists of two distinct phases: the particulate phase and the gaseous phase (Maga, 1988). The particulate phase accounts for about 90% of the total volume, and contains many undesirable or unnecessary components which are not important in fish smoking. The 10% gaseous phase contains most of the desirable components, and is essential for good smoke color and flavor. Over 390 different compounds have been detected in wood smoke, and it has been estimated that smoke may contain over 1,000 compounds. Of these, over 70 compounds have been isolated in smoked foods.

For both liquid and traditional smoke, the compounds most important to fish smoking are divided into three basic categories: phenols, carbonyls, and acids. These compounds are mostly contained in the gaseous phase of traditional smoke and are responsible for various desirable and undesirable effects (Maga, 1988).

Why Do We Smoke Fish? The desirable effects of the smoking process include its influence on color, flavor, aroma, antioxidant activity, and anti microbial activity (Horner, 1992). These desirable effects are common to both traditional smoke and liquid smoke.

Smoke Color Smoke absorption, heating, and drying all play a role in smoked color development. The smoked color of fish products is developed when the carbonyl in the smoke are absorbed into the moist product surface, and then react with the amino groups in fish muscle to form the desired smoked color (Horner, 1992). The carbonyl compounds are mainly responsible for smoked color development, although the phenol compounds also contribute to color. The color reaction between the carbonyl from the smoke and the amines in the fish is enhanced by hot, dry conditions. Even if the product is not smoked, some surface browning will still occur from the reaction of amino groups with carbonyl that are naturally available in fish. This is why products that are exposed to dry heat will develop a brown surface color even when smoke is not applied. The basic mechanism of smoke color formation is the same as that of a reaction known as the Maillard browning reaction (Horner, 1992). Therefore, as with Maillard browning, color formation is promoted by hot, dry conditions. This is why, after liquid smoke is applied, a hot, dry 'color setting' step is used to develop and set the smoke color before proceeded with humidified cooking steps.

Flavor & Aroma Smoked fish flavor and aroma is mainly due to the phenols present in wood smoke (Horner 1993). Over 20 different phenols contribute to smoked flavor, some of which also contribute to the aroma of smoked fish (Maga 1988).

Antioxidant Activity Smoke has a strong antioxidant activity that is associated with the phenol groups (Horner 1992; Maga 1988). The antioxidant activity helps to prevent development of off flavors due to oxidative rancidity. This is especially important to fish processors since fish are prone to oxidative rancidity.

Anti Microbial Activity The smoking process kills bacteria and prevents their growth because of the combined effects of heating, drying, reduced pH, and anti microbial smoke components (Horner 1992; Maga 1988). The smoke components that are primarily responsible for the anti microbial effect are the phenol compounds that actually kill bacteria. The acids in smoke are responsible for the reduced pH.

Undesirable Effects Along with the desirable effects of smoking, there are also some undesirable effects including the degradation of some nutrients and contamination of fish products with potentially toxic compounds from the smoke. The nutrient degradation is of minor importance, and will not be further discussed.

Contamination with carcinogenic compounds The most prominent undesirable effect of smoking is the potential for product contamination with polycyclic aromatic hydrocarbons (PAH) that exist in wood smoke. PAH compounds such as benzopyrene have been demonstrated to be mutagenic and carcinogenic in numerous animal experiments (Horner, 1992; Maga, 1988; Sikorski, 1988). Researchers have found that the contamination of product by benzopyrene is related to very high smoke generation temperatures. As a result, smoke generation temperatures of less than 750° have been recommended to reduce PAH contamination of wood smoked products (Regenstein and Regenstein, 1991). PAH compounds are removed during the production of liquid smoke, and therefore the contamination of liquid smoked fish products with PAH compounds is not a concern (Regenstein and Regenstein, 1991).

Traditional Wood Smoke Application Traditional wood smoke used in smokehouses is produced using several different types of smoke generators. Many factors influence the effectiveness of wood smoke application, including fish surface moisture content, oven temperature, relative humidity, air velocity, and smoke generator temperature. The most important of these variables is the amount of moisture available to absorb smoke on the fish surface during smoking.

Surface Moisture Research has shown that smoke is absorbed more rapidly by fish muscle with a high surface moisture content than with a low surface moisture content. Foster et al. (1961) found that the rate of phenol absorption by pre-dried fish was only 5% of that absorbed by wet fish.

Figure 4 shows that when fish having different moisture contents were smoked, the phenol absorption was higher for the fish with higher moisture content than for the drier product. In another study on hot smoked fish, researchers found that smoke absorption decreased as the product surface dried (Figure 5). As shown in Figure 5, fish that were smoked for 2 hours, more smoke was absorbed in the first 20 minutes than in the remaining 100 minutes. This suggests that if you want a darker smoked color, you don't necessarily have to smoke longer, just start smoking sooner.

This research shows that adequate surface moisture is necessary for smoke absorption, and that increased surface moisture levels will enhance smoke absorption. However, what would happen if you put a load of salmon fillets in the oven and started smoking right from the get-go? The wet surface fillets would readily absorb the smoke, resulting in an undesirable dark brown color. Even though adequate surface moisture is necessary for smoke absorption, keep in mind that too much surface moisture during smoking may result in an excessively dark black or brown smoked color, and may even cause streaking.

Initial Surface Conditions If the initial surface moisture conditions for fish vary from load to load, these uneven surface conditions will result in inconsistent smoked color from load to load.

Figure 6 shows the effect of uneven surface conditions on smoke absorption by fish fillets (Toledo, 1993). Fillets that were allowed to dry at room temperature for six hours before smoking absorbed the least smoke. Control fillets that were brined and held in a holding cooler for 16 hours before smoking absorbed the most smoke. Product that was steamed for two minutes prior to smoking absorbed almost as much smoke as the control.

As shown in Figure 6, variation in product surface conditions, such as between fish that has staged in a cooler versus staged at room temperature, can have a strong effect on how much smoke the product absorbs. To reduce any variation in surface moisture within a load or between loads, a conditioning step can be used to uniformly moisten all of the product surfaces before moving to the drying step. This conditioning step is the first step in the process, and is usually a short, low temperature, high humidity step. A commonly used conditioning step is 15 minutes at 100° dry bulb with the dampers closed.

Smokehouse Airflow The effect of airflow on smoke absorption is shown in Figure 7 (Toledo, 1993). This diagram compares the amount of smoke absorbed for fillets smoked in either a free or a forced convection smokehouse. As shown in Figure 7, the fillets smoked with fan driven air absorbed much more smoke than those smoked in static air.

Liquid Smoke Application Many fish processors have successfully replaced traditional smoking of fish products with the application of liquid smoke. Liquid smoke contains the same functional components (phenols, carbonyl, and acids) that are found in traditional smoke. However, an advantage of using liquid smoke is that the undesirable PAH compounds such as benzopyrene has been removed. Another advantage of liquid smoke is that the emission of undesirable pollutants from the oven to the atmosphere is reduced.

For fish products, liquid smoke is most commonly applied by atomizing a cloud of liquid smoke into the oven. Atomization is the most widely used application system because it is easily adapted to batch oven installations. Product that has been liquid smoked using atomization may be labeled as 'smoked'.

The most common procedure for applying liquid smoke is the atomize/dwell method. Using this procedure, the product is usually dried for a period of time before smoking. The oven is then shut down and a pre-measured amount of liquid smoke is atomized into the oven. After atomization is completed, the atomized cloud is usually allowed to dwell or 'rest' for preset time period before the oven is restarted and the process continued. This atomize/dwell procedure is sometimes repeated two or three times during the process. If it takes longer than 30 minutes to atomize the required amount of smoke into the oven, then it may be a good idea to break the atomization step into two or more steps. For example, if it takes 50 minutes to atomize the required amount of liquid smoke into an oven for a full load, it would be more effective to use two 25 minute atomization steps separated by a 30 minute drying step. As with any other type of smoke application, a period of dry heat is necessary after smoke application for proper color development.

Any color uniformity problems that occur using the atomization method can sometimes be resolved by shortening the dwell time or by running the main blower to circulate the atomized cloud during the last part of the dwell time. Regardless of the atomization time, a dwell time of longer than 5 to 10 minutes is probably unnecessary, and dwell times longer than 10 minutes often cause undesired color variation within the oven.

Another method of atomization is to continuously atomize liquid smoke into the oven air stream during one or more smoke cook steps, without shutting down the oven. Using this method, the atomized liquid smoke is treated much like a traditional wood smoke process. Low liquid smoke flow rates are necessary to avoid excessive consumption of smoke.

Smoked Color Development and Uniformity

Most smoked color problems fall into two basic categories: -Color development problems -Color uniformity problems.

Either you're not getting the color you want, or you're getting the color that you want, but not on all of the product.

Almost any question as to how to improve smoked color development or uniformity will usually result in a tidal wave of often contradictory advice. I will try not to add to that tidal wave of contradictions here, but rather offer the following simple suggestions to keep in mind when developing or revising smoking processes.

-Make sure that your smoke generator or liquid smoke application system is working properly and consistently from day to day. This might seem obvious, but it is difficult to develop a good, uniform smoked color with inadequate or inconsistent levels of smoke.

-Start with uniform product surface conditions. Variation in product surface conditions must be minimized as much as possible before smoking. For example, if some loads go right from brine into the oven, and others are allowed to stage in a cooler or at room temperature for some time before being moved into the oven, these different surface moisture conditions will almost certainly result in uneven smoked color from load to load. Uneven surface conditions can be evened out by using either a lengthened drying step or a short, low temperature, high humidity conditioning step (for example: 15 minutes at 100°F dry bulb, dampers closed). Non-uniform surface conditions before smoking will result in non-uniform smoked color after smoking.

-When developing or revising smoking schedules, test the effect of a shorter drying time by holding out a few test pieces from the production load and putting them back in after the load has dried for a pre-set time period. The proper surface moisture content for smoking will vary greatly depending on the product type. In general, a wetter fish surface before smoking will result in a darker smoked color, while a drier surface will result in a lighter smoked color. If you want a darker smoked color, you don't necessarily have to smoke longer, just start smoking sooner. However, if the product surface is not adequately dry when the smoke is applied, the product may develop an undesirable dark brown or even black color.

-Along with the effect of drying time on smoked color, don't forget the importance of drying the fish to improve firmness (especially for whole fish), develop the pellicle, and prevent curd formation.

Conclusion Of all the many processes involved in the manufacture of smoked fish products, the smoking process itself is probably one of the least well understood. As such, fish smoking remains more of an art than a science. However, if the basic principles of cooking and smoking described in this paper are understood and followed, it will make this easier for you to successfully develop and improve your smoking processes, and thus enhance the quality and value of your smoked fish products to your customers.