Добавил:
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:

The Physics of Coronory Blood Flow - M. Zamir

.pdf
Скачиваний:
157
Добавлен:
10.08.2013
Размер:
5.78 Mб
Скачать

330 9 Basic Unlumped Models

a tube segment within a tree structure, any departure of the pressure distribution from the “benchmark” of a uniform normalized value of 1.0 provides a direct and useful measure of the e ects of wave reflections in that structure. It is also a highly meaningful measure because the pressure distribution produced by wave reflections in a vessel segment within a vascular tree is superimposed on a pre-existing pressure distribution associated with steady flow within that vessel segment, and the net result is a modified distribution of pressure gradients driving the flow. For the same reasons, the pressure distribution along a vascular tree as a whole has the same utility and functional significance. The focus in this section, therefore, is on determining the pressure distribution within a vascular tree structure under di erent conditions.

We recall from Section 8.7 that an oscillatory input pressure applied at entry to an elastic tube produces a travelling wave within the tube. If this scenario occurs in a tube segment within a vascular tree structure, the travelling wave will progress to the next tube segment within the hierarchy of the tree structure, with the junction point between the two segments usually acting as a wave reflection site because of any di erence in the admittance properties of the two tubes. Typically, only part of the wave will be reflected. The remaining part, usually referred to as the “transmitted” part of the wave, will continue on to the next junction point, and so on. The net result is that the pressure distribution within the tree structure will be modified because of the many reflected waves within the system.

In order to determine the pressure distribution in a branching tree structure, it is necessary to keep track of the “net” pressure wave as it progresses from the root segment of the tree to the periphery, the net pressure wave in each tube segment being the sum of the transmitted wave and any reflected waves. In each tube segment along the way, we use essentially the same analysis as that in Section 8.7 for wave reflections in a single tube, but now the tube position within the tree structure must be identified so that the pressure and flow within it can be mapped along with the pressure and flow within the tree as a whole. For this purpose we use the j, k notation introduced in Section 9.2, whereby j denotes the level of the tree in which a given tube segment is located and k denotes the sequential position of that segment within other segments at that level, as illustrated in Figs. 9.2.3, 4.

Using that notation, the results of Section 8.7 for wave reflections in a single tube can now be placed in the context of a tree structure. In particular, the net pressure wave within a general tube segment at position j, k in a tree structure, using Eqs.8.7.3,19, is now given by

Pj,k (xj,k , t) = p0,j,k e(txj,k /cj,k ) + Rj,k e(t(2lj,k xj,k )/cj,k ) (9.6.1)

where lj,k is the length of the tube segment and xj,k is a position coordinate within that length such that the tube segment extends from xj,k = 0 to xj,k = lj,k , Rj,k is the reflection coe cient at the downstream end of the vessel segment, cj,k is the wave speed and p0,j,k is the amplitude of the input

9.6 Pulsatile Flow in Elastic Branching Tubes

331

oscillatory pressure at entry to that segment, namely

pin,j,k (t) = p0,j,k eiωt

(9.6.2)

As was done in Section 8.7, the net pressure wave in Eq. 9.6.1 can now be broken into its oscillatory components in time and space such that

Pj,k (xj,k , t) = pj,k (x) × eiωt

(9.6.3)

where

 

pj,k (x) = p0,j,k eiωxj,k /cj,k + Rj,k e(2lj,k xj,k )/cj,k

(9.6.4)

The absolute value of this complex function of x, namely |pj,k (x)|, is what we refer to as the “pressure distribution” in a tube segment within a tree structure, and the map of all such distributions together make up the pressure distribution in the tree as a whole. It is a key property of the propagating pressure wave within each tube segment. It represents the net pressure at a fixed point in time at each point along the segment, including forward and backward moving waves. Also, in view of Eq. 9.6.3, it represents the amplitude of time oscillations in pressure at a fixed position within the tube segment.

In the context of a tree structure, it is seen from Eq. 9.6.4 that the pressure distribution in a vessel segment at position j, k within the tree depends on the wave speed cj,k within that segment, on the reflection coe cient Rj,k at the downstream end of the segment, and on the input pressure amplitude p0,j,k at entry to the tube segment. The way in which each of these is determined is outlined below.

Depending on the desired accuracy, the wave speed in a vessel segment within a tree structure may be taken as the constant Moen-Korteweg wave speed c0 as defined by Eq. 8.6.1 and which depends on properties of that segment only, or it may be taken as the wave speed c as defined by Eq. 8.6.2 based on a solution of pulsatile flow in an elastic tube which depends on properties of the tube as well as on the frequency parameter Ω. For comparison purposes, results based on these two alternatives to be presented below shall be identified by c = c0 and c = c respectively.

The reflection coe cient at the downstream end of a tube segment in a tree structure is determined by the characteristic admittance of that segment and by the e ective admittances of the two branch segments forming the bifurcation at that end, as discussed in Section 9.5. If the position of the vessel segment under consideration is j, k, then the positions of the two branch segments at its downstream end (xj,k = lj,k ) are j + 1, 2k − 1 and j + 1, 2k, as illustrated in Figs. 9.2.3, 4. Using the results of Section 9.5, therefore, Eq. 9.5.20 expressed in the notation of the present section gives for the reflection coe cient

Rj,k =

Y0,j,k (Ye,j+1,2k−1

+ Ye,j+1,2k )

(9.6.5)

+ Ye,j+1,2k )

 

Y0,j,k + (Ye,j+1,2k−1

 

332 9 Basic Unlumped Models

The characteristic admittance Y0 is defined by Eq. 9.5.5 and is determined by the radius of the tube segment and by the wave speed which again may be taken as c or c0 as discussed above, depending on the desired accuracy. The e ective admittance Ye, using Eq. 9.5.22 in present notation, is given by

Ye,j,k = Y0,j,k ×

Y0,j,k + i(Ye,j+1,2k−1

+ Ye,j+1,2k ) tan θj,k

 

 

 

 

(Ye,j+1,2k−1

+ Ye,j+1,2k ) + iY0,j,k tan θj,k

 

θj,k =

ωlj,k

 

 

 

 

(9.6.6)

cj,k

 

 

 

 

 

 

 

 

 

Finally, the input pressure amplitude p0,j,k at entry to the tube segment at position j, k is obtained by equating the local pressure in all three vessel segments that meet at that point (xj,k = 0), recalling that in a branching tree structure based on repeated bifurcations, three vessel segments meet at entry to and at exit from each interior tube segment, as illustrated in Figs. 9.2.3, 4. For junctions beyond the root level of the tree, that is for j > 0, this gives

p0,j,k (x) = p0,j−1,n

(1 + Rj−1,n)eiωlj−1,n /cj−1,n

,

j > 0 (9.6.7)

 

1 + Rj,k ei2ωlj,k /cj,k

where

 

 

 

 

 

 

 

n =

k

if k is even

 

(9.6.8)

 

2

 

 

 

 

 

 

 

 

=

k + 1

if k is odd

 

(9.6.9)

 

 

2

 

 

 

 

 

 

 

 

 

 

For the root segment of the tree where j = 0 the input pressure amplitude p0,0,1 is clearly equal to the amplitude of input pressure to the tree as a whole. That is, if the latter is given by

p(t) = p0eiωt

(9.6.10)

then

 

p0,0,1 = p0

(9.6.11)

Note, however, that the oscillatory pressure that actually prevails at the root of the tree is given by

P0,1(0, t) = p0,0,1

1 + R0,1e−i2θ0,1

eiωt

(9.6.12)

 

 

i2θ0,1

 

 

 

= 1 + R0,1e

p0e

iωt

(9.6.13)

 

 

= p(t)

unless R0,1 = 0

 

 

(9.6.14)

and

9.6 Pulsatile Flow in Elastic Branching Tubes

333

|P0,1(0, t)| = |p0(t)|

(9.6.15)

= p0 unless R0,1 = 0

(9.6.16)

In other words, the oscillatory pressure that actually prevails at entry to the tree is in general not equal to the input oscillatory pressure, and the amplitude of the oscillatory pressure at entry to the tree is in general not equal to the amplitude of the input pressure. The di erence in each case is due to wave reflections from downstream junctions.

To illustrate these results, examples of pressure distributions are presented below for the 5-level and the 11-level tree models used earlier. In each case the two bounding paths are identified as path-1 and path-2, where the first path follows the branch with the larger diameter at each bifurcation while the second path follows the branch with the smaller diameter at each bifurcation, as illustrated in Fig. 9.6.1.

path-1 path-2

path-1 path-2

Fig. 9.6.1. The 5-level and the 11-level tree models along which the pressure distributions are calculated to illstrate the e ects of wave reflections. The bounding paths identified as path-1 and path-2 are of particular interest because they represent the paths along which the diameter of vessel segments diminish most slowly (path-1) or most rapidly (path-2).

Pressure distributions along the 5-level tree are shown in Figs.9.6.2-4 at high, moderate, and low frequencies, respectively. We recall that the pressure

334 9 Basic Unlumped Models

normalized pressure

 

 

 

 

f = 10 Hz

 

 

 

 

 

c=c0

 

1

 

 

 

γ = 3

 

 

 

 

 

α = 0.7

 

0.8

 

 

 

 

 

 

 

 

path−2

 

 

0.6

 

 

path−1

 

 

 

 

 

 

 

0.4

 

 

 

 

 

0.2

 

 

 

 

 

00

1

2

3

4

5

 

 

 

level

 

 

Fig. 9.6.2. Pressure distribution along di erent paths within the 5-level tree model, at a frequency of 10 Hz. In the absence of wave reflections the pressure distribution would be uniform at a normalized constant value of 1.0, thus the observed deviations from this benchmark are due entirely to the e ects of wave reflections. Pressure distributions along the two bounding paths defined in Fig. 9.6.1 are shown in bold. They are seen to be “bounding” here too, in the sense that they represent bounding values for the pressure distribution in the tree as a whole.

shown is normalized such that in the absence of wave reflections the pressure distribution would be uniform at a normalized value of 1.0 (Eq. 8.7.14). The figures show that in the presence of wave reflections the pressure distributions lie below this value at all three frequencies. At the highest frequency, which corresponds to the lowest wave length, the pressure decreases from the root segment of the tree to the periphery, while at the lowest frequency the pressure is uniformly below the normalized benchmark value. This is consistent with e ects of wave reflection at high and low wave-length-to-tube-length ratio discussed in Section 8.7.

Corresponding results for the 11-level tree model are shown in Figs.9.6.5- 7. The results indicate that the e ects of wave reflection are even more pronounced here than in the 5-level tree model, as indicated by larger deviations of the pressure distribution from the benchmark of uniform distribution at a normalized value of 1.0. This indicates clearly that the e ects of wave reflections in a vascular tree structure are cumulative, thus producing larger overall e ects in the 11-level tree where there are a larger number of wave reflection sites.

The dimensions of the 11-level tree model on which the results in Figs.9.6.2- 4 are based have been chosen to be of the same order of magnitude as those of the human coronary arterial tree, with the root segment of the tree having

9.6 Pulsatile Flow in Elastic Branching Tubes

335

a diameter of 4.0 mm and the diameters of its subsequent branches diminishing hierarchically according to the cube law thereafter. Results in Section 9.4 showed that these dimensions produce fairly high wave-length-to-tube-length ratios, in excess of 200 throughout the 11-level tree. It has been thought in the past that because such high ratios lead to what seem to be “inconspicuous” e ects in a single tube (Fig. 8.7.9 compared with Fig. 8.7.4), the same would likely be true in the coronary circulation. The results in Figs.9.6.5-7 indicate clearly that this is not the case. The reasons for this are twofold. First, while the e ects of wave reflections in a single tube at high wave-length-to-tube- length ratio are inconspicuous, they are not insignificant. As seen clearly in Fig. 8.7.9, while the pressure distribution within the tube remains uniform under these circumstances, it is so at a much higher level than the benchmark level of 1.0 when wave reflections are absent. Second, in a branching tree consisting of many tube segments these e ects are cumulative because of the large number of reflection sites.

To pursue these issues further, the dimensions of the 11-level tree can now be changed to the e ect that they become of the order of magnitude of the human systemic arterial tree rather than the coronary arterial tree. To do so we give the root segment of the tree a diameter of 25 mm, being representative of the human aorta, and let subsequent branches assume diminishing diameters in accordance with the cube law as before. The results are shown in Figs.9.6.7-9. The e ects of wave reflections on the pressure distribution are indeed more conspicuous here and have the characteristic highs and lows associated with low wave-length-to-tube-length ratios (as in Figs. 8.7.5, 6). The di erence between these results and those based on the smaller coronary scale

normalized pressure

 

f = 5 Hz

 

c=c0

1

γ = 3

 

α = 0.7

0.8

 

 

path−2

0.6

path−1

 

0.4

0.2

00

1

2

3

4

5

 

 

 

level

 

 

Fig. 9.6.3. Pressure distribution along di erent paths within the 5-level tree model, as in Fig. 9.6.2, but here at a frequency of 5 Hz.

336 9 Basic Unlumped Models

normalized pressure

 

f = 1 Hz

 

c=c0

1

γ = 3

 

α = 0.7

0.8

 

 

path−2

0.6

path−1

 

0.4

0.2

00

1

2

3

4

5

 

 

 

level

 

 

Fig. 9.6.4. Pressure distribution along di erent paths within the 5-level tree model, as in Fig. 9.6.2, but here at a frequency of 1 Hz.

(Figs. 9.6.5–7) is a di erence in the pattern of pressure distributions which they produce rather than a di erence of significance. While under the smaller scale of the coronary vasculature the pressure distributions have a monotonically decreasing pattern at all three frequencies, under the larger scale of the systemic vasculature there are regions of increasing pressure at the higher frequencies and at levels of the tree closer to the root segment. It is under these conditions that the well known “peaking” phenomenon occurs in the systemic circulation, but not in the coronary circulation where these conditions are absent.

All the results above are based on the wave speed approximation c = c0 as indicated in the figures, where c0 is the constant Moen-Korteweg wave speed. Results based on the more accurate c = c model, where c is the wave speed obtained from a solution of pulsatile flow in an elastic tube, are shown in Figs.9.6.11-13. Comparison of these with results in Figs.9.6.5-7 based on c = c0 indicates that the di erence between the two is not highly significant.

A more significant di erence arises if the hierarchy of branch diameters within the tree structure is changed from one based on the cube law (γ = 3) to one based on the square law (γ = 2). Under the square law the sum of the cross-sectional areas of the two branch segments is equal to the cross-sectional area of the parent segment at all bifurcations within the tree structure. If this is combined with the c = c0 wave speed assumption, then there is no change of admittance as the flow crosses each bifurcation within the entire tree structure, which means that the bifurcations are no longer reflection sites

9.6 Pulsatile Flow in Elastic Branching Tubes

337

 

 

 

 

 

 

f = 10 Hz

 

 

 

 

 

 

c=c0

 

1

 

 

 

 

γ = 3

 

 

 

 

 

 

α = 0.7

pressure

0.8

 

 

 

 

 

 

 

 

 

 

 

normalized

0.6

 

 

 

 

 

0.4

 

 

 

path−2

 

 

 

 

 

 

 

 

 

 

 

 

0.2

 

 

 

path−1

 

 

 

 

 

 

00

2

4

6

8

10

 

 

 

 

level

 

 

Fig. 9.6.5. Pressure distributions along di erent paths within the 11-level tree model at a frequency of 10 Hz. Comparison with the corresponding results for the 5-level tree model shown in Fig. 9.6.2 indicates clearly that the e ects of wave reflections are larger here, as marked by larger departures from the benchmark of uniform distribution at a normalized value of 1.0 which occurs when wave reflections are absent.

 

 

 

 

 

 

f = 5 Hz

 

 

 

 

 

 

c=c0

 

1

 

 

 

 

γ = 3

 

 

 

 

 

 

α = 0.7

pressure

0.8

 

 

 

 

 

 

 

 

 

 

 

normalized

0.6

 

 

 

 

 

0.4

 

 

 

path−2

 

 

 

 

 

 

 

0.2

 

 

 

path−1

 

 

 

 

 

 

00

2

4

6

8

10

 

 

 

 

level

 

 

Fig. 9.6.6. Pressure distributions along di erent paths within the 11-level tree model as in Fig. 9.6.5 but here at a frequency of 5 Hz.

338

9 Basic Unlumped Models

 

 

 

 

 

 

 

 

 

 

 

f = 1 Hz

 

 

 

 

 

 

 

c=c0

 

 

1

 

 

 

 

γ = 3

 

pressure

 

 

 

 

 

α = 0.7

 

0.8

 

 

 

 

 

 

normalized

 

 

 

 

 

 

0.6

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.4

 

 

 

 

 

 

 

 

 

 

 

path−2

 

 

0.2

 

 

 

path−1

 

 

00

2

4

6

8

10

 

 

 

 

 

level

 

 

Fig. 9.6.7. Pressure distributions along di erent paths within the 11-level tree model as in Fig. 9.6.5 but here at a frequency of 1 Hz.

 

1.2

 

 

 

 

f = 10 Hz

 

 

 

 

 

 

 

 

 

 

 

 

c=c0

 

1

 

 

 

 

γ = 3

pressure

 

path−2

 

 

 

α = 0.7

0.8

 

 

 

 

 

 

path−1

 

 

 

 

normalized

0.6

 

 

 

 

 

0.4

 

 

 

 

 

 

 

 

 

 

 

 

0.2

systemic scale

 

 

 

 

 

00

2

4

6

8

10

 

 

 

 

level

 

 

Fig. 9.6.8. Pressure distributions along di erent paths within the 11-level tree model at a frequency of 10 Hz as in Fig. 9.6.5 but here the root segment of the tree has a diameter of 25 mm compared with 4 mm in that figure. Thus the scale of the tree model in this case is representative of the scale of the human systemic arterial tree while that in Fig. 9.6.5 is representative of the coronary arterial tree. Wave-length-to-tube-length ratios are much lower in this case, thus the characteristic highs and lows in the pressure distributions.

9.6 Pulsatile Flow in Elastic Branching Tubes

339

 

1.2

 

 

 

 

f = 5 Hz

 

 

 

 

 

 

 

 

 

 

 

 

c=c0

 

1

 

 

 

 

γ = 3

pressure

 

 

 

 

 

α = 0.7

0.8

 

 

path−1

 

 

 

 

 

 

 

 

 

 

 

 

 

normalized

0.6

path−2

 

 

 

 

 

 

 

 

 

0.4

 

 

 

 

 

 

 

 

 

 

 

 

0.2

systemic scale

 

 

 

 

 

00

2

4

6

8

10

 

 

 

 

level

 

 

Fig. 9.6.9. Pressure distributions along di erent paths within the 11-level tree model as in Fig. 9.6.8, but here at a frequency of 5 Hz which means higher wave lengths and thus higher wave-length-to-tube-length ratios than in that figure.

 

1.2

 

 

 

 

f = 1 Hz

 

 

 

 

 

 

 

 

 

 

 

 

c=c0

 

1

 

 

 

 

γ = 3

pressure

 

 

 

 

 

α = 0.7

0.8

 

 

 

 

 

 

 

 

 

 

 

normalized

0.6

 

 

 

 

 

0.4

 

 

 

 

path−2

 

 

 

 

 

 

 

0.2

systemic scale

 

 

 

path−1

 

 

 

 

 

 

 

00

2

4

6

8

10

 

 

 

 

level

 

 

Fig. 9.6.10. Pressure distributions along di erent paths within the 11-level tree model as in Fig. 9.6.8, but here at a frequency of 1 Hz, which means considerably higher wave lengths and thus considerably higher wave-length-to-tube-length ratios than in that figure.