Phase diagrams reveal acid-gas injection subtleties

H₂S INJECTION-1

John J. Carroll
Gas Liquids Engineering
Calgary

Examination of phase equilibria in a binary hydrogen sulfide (H ₂S) and water system explains some nuances of acid-gas injection.

Also, the phase diagrams dispel some misconceptions, prevalent in literature, on hydrate formation.

This article shows how hydrates can form without free water (an aqueous liquid phase) present and how hydrates can also form in the presence of a nonaqueous liquid.

This first of a two-part series describes multicomponent phase diagrams. The concluding part will explain acid-gas phase changes from the amine unit to the bottom of an injection well.

Phase diagrams

Phase diagrams illustrate some fascinating and complex behavior encountered in the binary H ₂S and water system. The design engineer should be aware that many phases might be encountered depending upon the temperature, pressure, and composition.

The engineer should question the phase effect on the process under consideration. Simple assumptions may lead to disastrous consequences.

The pressure-vs.-composition (P-x) diagrams show that the amount of water present in a mixture plays an important role in determining whether or not a hydrate will form. If there is too much water, all the gas will dissolve in the liquid and no hydrate forms. If there is too little water, the water is dissolved in the gas (or liquid) and again a hydrate does not form.

Thus, just because temperature and pressure are correct for forming hydrates does not mean that a hydrate will form.

One myth about hydrate formation is that the presence of free-water phase is required. But phase diagrams show that a hydrate can form without free water being present. This is a potential problem in acid-gas injection.

Even though the gas seems dry, so dry that an aqueous phase does not form, hydrate formation is still possible.

Hydrates can coexist with a nonaqueous liquid. Just because a nonaqueous liquid is present does not mean that a hydrate cannot form.

CO₂ plus water and H₂S plus water systems have similar phase equilibria, except that the phenomena occur at different pressures and temperatures. This is significant because CO₂ is the other significant component in acid gas.

Because acid gas does not contain only H₂S and water, the design engineer should be aware of the implications of a multicomponent mixture that are encountered in actual practice.

Acid gas

In the past, gas plants producing only a small amount of acid gas (typically less than 1 metric ton of sulfur/day) were permitted to flare the gas. However, because of environmental concerns, flaring is becoming less common.

To replace flaring, acid-gas injection is quickly becoming an environmentally friendly alternative.

Acid-gas injection also has attracted the attention of larger producers. With sulfur prices depressed, reinjection could be an economical alternative to building a Claus plant and stockpiling sulfur.

It is even possible that injected H₂S could be produced at a future date if sulfur prices improve.

The injected acid-gas stream coming off of the amine regeneration tower typically contains a mixture of H₂S and CO₂, light hydrocarbons (mostly methane), and some nitrogen. It typically is saturated with water, and water is the key to designing the injection scheme.

The presence of water poses several problems, not the least of which are the formation of a corrosive aqueous phase and the formation of hydrates, which can plug lines and equipment.

Phase behavior

To construct an H ₂S and water pressure-vs.-temperature (P-T) diagram, the pure components must be examined first. The vapor pressure curves for the two components are well known. Other phase boundaries for the pure components have also been established.

Next, one must consider the binary mixture. The various three-phase loci involving solids, for temperatures greater than about 0° C., have been studied often, and a thorough review was presented in Reference 1.

Also, there have been several fixed points established for both the pure components and the binary mixture. Table 1 [51,726 bytes] summarizes the known fixed points for the binary H₂S and water system.

The three-phase locus involving fluid phases (L_A + L_S + V), including the three-phase critical end-point, was discussed in Reference 2. The nomenclature is as follows:

• L_A-Aqueous liquid (water-rich) liquid, sometimes referred to as free water
• L_S-H₂S-rich liquid
• V-Vapor.

A small critical locus extends from the critical point of pure H ₂S to the three-phase critical end-point. Another critical locus extends from the critical point of water. This region has not been studied, but it is speculated that this critical locus extends to higher temperatures and lower pressures.

From this information, Carroll and Mather¹ constructed a (P-T) diagram for this system. Fig. 1 [84,853 bytes] is a similar figure, covering a smaller range of temperatures.

Fig. 1 shows:

• The pure component two-phase loci (notably the vapor pressure curves)-the broken lines
• The binary three-phase loci-the solid lines
• The binary critical loci-the dotted curves.

It is important to note that this figure is not to scale.

Fig. 1 is the two-dimensional projection of a three-dimensional situation. The composition axis has been eliminated by this projection. The P-x diagrams in Fig. 2 [91,509 bytes] and Fig. 3 [198,013 bytes] show how the composition affects the phase equilibria.

A quadruple point, a point where four phases are in equilibrium, occurs at the intersection of four three-phase loci. There are two quadruple points shown in Fig. 1.

The first of these, Q₁, appears to also intersect the vapor pressure of H₂S. In reality it does not. It is at the same pressure and temperature as the vapor pressure (to within experimental error), but remember, this plot is a projection.

The vapor pressure curve and the Q₁ point are separated in the third dimension, composition, which is not clearly indicated.

To be convinced of this, one need only consider that the vapor pressure is a single-component phenomenon. This is only for pure H₂S. A quadruple point can only exist for binary and multicomponent systems. Thus, water must be present for the quadruple point to form.

One interesting result arising from the observation is that Point Q₁ and the vapor pressure of H₂S are at the same temperature and pressure. The effect of this is that the L_S + H + V locus lies above the vapor pressure, whereas the L_S + L_A + V lies below. H is the hydrate.

This crossing of the three-phase loci and the vapor pressure has an interesting consequence. Furthermore, although they are clearly separated in the diagram, these two loci and the vapor pressure of H₂S are almost coincident.

In addition, there are other loci that appear to intersect in the P-T diagram. This is usually not the case. For example, the L_S + H + V and the L_A + H + I loci do not intersect (I is ice). The loci only intersect at pure component triple points and binary quadruple points. These other apparent intersections are due to the elimination of the composition axis, as mentioned previously.

Phase diagram at 15° C.

The temperature of 15° C. lies between the triple point of water and the first quadruple point. Starting at a low pressure and going to high pressures in Fig. 1, we can see that the loci intersected are:

• Vapor pressure of water (L_A + V)
• Aqueous liquid + hydrate + vapor (L_A + H + V)
• Vapor pressure of H₂S (L_S + V)
• H₂S-rich liquid + hydrate + vapor (L_S + H + V).

These are used to construct the pressure-composition (P-x) diagram.

Fig. 2 shows the P-x diagram for the binary H₂S and water system at 15° C. The single-phase and two-phase regions are labeled, and the three-phase points are the horizontal lines. Note, this plot is not to scale.

The pressures for the three-phase points and the vapor pressures of the pure components are noted on the plot. The equilibrium compositions for the three-phase L_A + H + V point are also shown.

For this, and subsequent P-x diagrams, many of the phase boundaries are shown as vertical lines (for example, constant composition). In reality, these boundaries would exhibit some slope and possibly some curvature.

These phase boundaries are not necessarily at constant composition. This poses an interesting problem, but has only a small potential effect on the discussion presented in the remainder of this article. It is, however, important that a design engineer be aware of this.

Consider the following experiment. A mixture of known composition is placed in an isothermal piston. The cylinder is transparent such that the number and nature of the phase can be clearly observed. The mixture is compressed by moving the piston, and thus the volume is reduced.

We could set this up in the lab and actually perform this experiment; however, after each time period in which we compress the mixture, we would have to wait for the mixture to equilibrate.

The following scenarios show the subtlety of the phase equilibria that the P-T diagram could never reveal.

Scenario 1

First, consider a mixture that is 99.5 mol % water. At a pressure below 1.7 kPa (the vapor pressure of water at 15° C.) the mixture exits as a single-phase vapor. In general, it is not true that a mixture must exist as a gas if the pressure is less than the vapor pressure of the least volatile component.

Many mixtures form azeoptropes; however, for the H₂S and water system, this is not the case.

As the mixture is compressed (such as when the volume in the piston is reduced), a point is reached where the first drop of aqueous phase forms. This is the aqueous dew point.

In Fig. 2, the aqueous dew point is represented by the lower curve.

At a fixed temperature, different mixtures have different dew-point pressures. For this mixture the dew point is at a pressure slightly greater than 1.7 kPa. The small amount of H₂S in the mixture has a small effect on the dew point, and thus it is almost the same as the vapor pressure of pure water.

As we continue to compress, the gas disappears and more liquid is formed. Throughout this liquefaction process, the pressure continues to rise.

This is unlike the liquefaction of a pure component, where the process is isobaric. In addition, throughout this compression the compositions of the phases change. Both the gas and liquid become richer in H₂S.

In the two-phase region the compositions of the two phases can be determined by drawing a horizontal line on the P-x diagram. The points where the line intersects the dew and bubble point loci give the composition of the two phases. The amounts of the phases present can be determined from the lever rule.

Eventually a point is reached where the entire mixture is liquefied. This is the bubble point.

Again, different mixtures have different bubble points. For this mixture the bubble point pressure is at about 215 kPa. However, as we will see, many mixtures do not have an aqueous phase bubble point.

Continued compression does not result in any further phase changes; the mixture remains a single-phase aqueous liquid.

Even though we are in a region where from the P-T diagram one would predict hydrate formation, no hydrate forms because there is insufficient H₂S present. All H₂S present remains in an aqueous solution.

Scenario 2

The second mixture we will consider contains 90 mol % water. Again at low pressure the mixture is single phase and is compressed until an aqueous dew point is reached. This dew point is at 1.9 kPa, a pressure slightly larger than the vapor pressure of pure water.

Continued compression results in forming more aqueous liquid, much the same as the first mixture. The relative amounts and composition of the phases can be determined in the manner discussed in Scenario 1.

However, in this case, a bubble point is not reached. Eventually, a point is reached where a solid just begins to form. This is the three-phase (L_A + H + V) point. This solid is a hydrate and the pressure at which this occurs is 464 kPa.

Once this point is reached, compressing the mixture does not change the pressure. What happens is that the relative amounts of the phases change, but not the pressure.

In this case, the vapor phase disappears and more solid and liquid are formed. Furthermore, throughout this three-phase compression the compositions of the three phases do not change. There is no simple method, such as the lever rule, that can be used to determine the amounts of the phases present from the P-x diagram.

Finally, we arrive at a point where all of the vapor disappears, a three-phase (L_A + H + V) bubble point.

Now compressing the mixture does result in a change of pressure, but not a change in phase. The mixture remains in two phases, L_A + H, regardless of the pressure. Since this is a two-phase region, the lever rule can be applied to determine the compositions and relative amounts of the phases.

In this scenario, there is sufficient H₂S present that, as one would speculate from the P-T diagram, a hydrate forms. Also, in this case the hydrate is formed in the presence of free water, the L_A phase.

Scenario 3

The third mixture is 50 mol % water. As with the first two mixtures, at low pressures the mixture is a gas that upon compression reaches an aqueous dew point. This dew point is at 3.4 kPa. Again, as with the first two mixtures, further compression results in more aqueous liquid being formed.

Once the pressure reaches the three-phase L_A + H + V point, the behavior is similar to Scenario 2. As with Scenario 2, this occurs at a pressure of 464 kPa. For a binary mixture a three-phase pressure, at a fixed temperature, is not a function of the composition. The compositions of the three phases are exactly the same as in Scenario 2.

Further compression does not change the pressure, only the relative amounts of the phases present. Unlike Scenario 2, where the vapor disappeared, in this case it is the aqueous liquid phase that disappears upon further compression. The point where the liquid is all gone is a three-phase (L_A + H + V) dew point.

Once the aqueous liquid phase is completely gone, the mixture enters a two-phase V + H region.

Upon further compression, the pressure increases, but the mixture remains in two phases until the three-phase L_S + H + V point is reached. At this point, an H₂S-rich liquid begins to form. This is a three-phase (L_S + H + V) dew point. This is at a pressure of 1,588 kPa, which is slightly greater than the vapor pressure of pure H₂S at this temperature (1,567 kPa).

Further compression results in changes in the amounts of the three phases but not in the pressure. In this respect, the behavior at this three-phase point (L_S + H + V) is similar to the three-phase point (L_A + H + V) encountered earlier.

Eventually, all of the vapor disappears. This is the three-phase (L_S + H + V) bubble point. At this point, the mixture is two phase, L_S + H.

Upon further compression, the pressure increases, but the mixture remains in two phases.

It is interesting to note that this mixture has two, three-phase bubble points, but the phases involved are different.

For this mixture, it is also interesting that a hydrate can exist in the presence of a nonaqueous liquid and without a free-water phase. In fact, the only place where free water and hydrate exist in equilibrium for this mixture is at a three-phase point.

However, believers in the "free-water theory" would say that a free-water phase was present during hydrate formation. However, this leads to the next scenario.

Scenario 4

This mixture contains 99.8 mol % H ₂S. Once again, at low pressure this mixture exists as a gas. This gas is compressed until the first drop of liquid forms. Unlike the previous mixtures, this phase is not an aqueous liquid but an H ₂S-rich liquid. This is the L _S dew point and is at a pressure slightly greater than the vapor pressure of pure H ₂S. This liquid is richer in H ₂S than the gas that it is in equilibrium with.

Further compression results in more liquid being formed and a reduction in the amount of vapor.

Eventually, the three-phase L_S + H + V is reached. This occurs at a pressure of 1,588 kPa. This is an incipient solid point. Surprisingly, even though this mixture is very lean in water, a hydrate forms.

Contrary to popular belief, a hydrate has formed and free water was never present. The L_A phase was never encountered in this scenario (and never will be regardless of the pressure) and yet a hydrate formed.

As with the other three-phase points, further compression does not result in a change in pressure, but a change in the relative amounts of the three phases.

In due time, all of the gas will disappear (a bubble point) and the two-phase L_S + H region is entered. Once again, in this region the pressure increases with further compression, but no further phase change occurs.

A point worth noting for this scenario (and also in the next scenario) is that a vapor phase can exist at pressure greater than the vapor pressure of the more volatile component. The presence of water, which is significantly less volatile then H₂S, makes the mixture more volatile.

Scenario 5

The final mixture is very rich in H ₂S, containing only a trace amount of water (for instance, 99.99% H ₂S). As with all of the previous mixtures, at low pressure the mixture exists as a single-phase vapor.

Similar to Scenario 4, this mixture is compressed until an L_S dew point is reached. This occurs at a pressure slightly greater than the vapor pressure of pure H₂S.

Further compression results in the formation of more H₂S-rich liquid, and the vapor disappears. Eventually, a point is reached where all of the vapor disappears. This is at the bubble point. The mixture then becomes single phase, which is an H₂S-rich liquid.

Additional compression beyond this bubble point does not result in further phase change. Note, this mixture does not contain enough water to form a hydrate, regardless of the pressure.

Phase diagram at 31° C.

Another interesting phase diagram for the H ₂S-water system is at 31° C., which is a temperature just slightly above the Q ₁ point. Fig. 3a shows the P-x diagram for this temperature.

One interesting aspect of this phase diagram is that it shows that this system can exhibit liquid-liquid immiscibility and form a hydrate.

Another interesting observation at this temperature is the nature of the formed H₂S-rich liquid. Unlike at 15° C., at 31° C. the L_S phase forms at a pressure less than the vapor pressure of H₂S.

In this case, the presence of water makes the mixture less volatile. Moreover, L_S is richer in water than the vapor. This too is the reverse of what occurred at 15° C. Both effects are a consequence of crossing the three-phase loci and the H₂S vapor pressure, which was discussed previously.

No experimental measurements of the H₂S hydrate have been made beyond about 33° C. However, if we extrapolate the correlation of Carroll and Mather1 to 100 MPa (14,500 psia), perhaps the upper limit of the pressure that interests the petroleum industry, it is estimated that the hydrate in this system occurs up to about 38° C. (100° F.).

Even at a temperature greater than the quadruple point, hydrates can still form.

Phase diagram at 50° C.

The phase diagram at 50° C. (Fig. 3b) is interesting from a compressor-design point of view. At this point, the design temperature for aerial coolers at compressor interstages is typically about 50° C.

The P-x diagram at 50° C. is similar to the 31° C. isotherm, except that regardless of the pressure, a hydrate is not encountered.

At this temperature, The L_S dew point is at a pressure slightly greater than the vapor pressure of pure H₂S. Furthermore, the L_S phase is richer in water than the equilibrium vapor.

The L_S + V region has been grossly exaggerated in this diagram. From the pressures shown in Fig. 3b, it extends only 50 kPa and is only about 2.5 mol % wide. It is a very small region that would take a very carefully designed experiment to examine it.

The behavior of a mixture very rich in H₂S is similar to the 31° C. isotherm, and the L_S phase is richer in water than the vapor.

Phase diagram at 103° C.

Another interesting phase diagram is for the 103° C. isotherm (Fig. 3c). This is a temperature greater than the critical point of pure H₂S, but less than the three-phase critical end point (the K point).

One fascinating feature is the occurrence of a binary critical point. At this temperature the critical point is about 9,150 kPa.

The composition of the critical mixture has never been measured experimentally. In Fig. 3c, the critical mixture is shown at a point intermediate to the compositions of the LS and V phases. It is equally likely that the critical composition is richer in H₂S than the three-phase vapor composition.

A second interesting property is that an H₂S-rich liquid can form at temperatures greater than the critical temperature of pure H₂S.

As with the P-x diagram at 50° C., the L_S + V region has been grossly distorted in Fig. 3c.

Phase diagram at 125° C.

At 125° C. (Fig. 3d), neither a second liquid nor a hydrate can form. This is merely gas solubility in a liquid. However, at this temperature, there is a maximum water content of the gas, which occurs at 9,250 kPa and about 95 mol % H ₂S.

An interesting consequence is that some mixtures have two dew points.

For example, a mixture containing 94 mol % H₂S is placed in the transparent cylinder discussed previously. Starting from low pressure (and hence a gaseous phase) the mixture is compressed. A point will be reached where the first drop of aqueous liquid appears. This is an LA dew point and it occurs at 5,380 kPa.

If the mixture is further compressed, initially the amount of liquid increases and the amount of vapor decreases. Eventually a point is reached where the amount of liquid reaches a maximum. Further compression results in a decrease in the amount of liquid. Such a counter intuitive action is called retrograde behavior.

After even further compression, a point is reached where all of the liquid disappears. This is a second dew point or a retrograde dew point. For this mixture, the retrograde dew point is at about 11,930 kPa.

Any additional compression results in the fluid remaining a single-phase gas.

On the other hand, mixtures rich in H₂S, those 95 mol % or greater, have no dew points and, therefore, will not form a liquid regardless of pressure.

References

1. Carroll, J.J., and Mather, A.E., Can. J. Chem. Eng., Vol. 69, 1991, pp. 1206-12.
2. Carroll, J.J., and Mather, A.E., Can. J. Chem. Eng., Vol. 67, 1989, pp. 468-70.

The Author