The arterial Windkessel
- 20k Downloads
Frank’s Windkessel model described the hemodynamics of the arterial system in terms of resistance and compliance. It explained aortic pressure decay in diastole, but fell short in systole. Therefore characteristic impedance was introduced as a third element of the Windkessel model. Characteristic impedance links the lumped Windkessel to transmission phenomena (e.g., wave travel). Windkessels are used as hydraulic load for isolated hearts and in studies of the entire circulation. Furthermore, they are used to estimate total arterial compliance from pressure and flow; several of these methods are reviewed. Windkessels describe the general features of the input impedance, with physiologically interpretable parameters. Since it is a lumped model it is not suitable for the assessment of spatially distributed phenomena and aspects of wave travel, but it is a simple and fairly accurate approximation of ventricular afterload.
KeywordsPressure-flow relation Arterial compliance Characteristic impedance Input impedance
Models are a simplification of reality which help to understand function. The arterial system has been modeled in many ways: lumped models [18, 73], tube models [8, 41, 80] and anatomically based distributed models [42, 64, 71]. In this paper we will discuss the lumped or Windkessel models. Lumped models of the venous system  will not be discussed.
The two-element Windkessel predicts that in diastole, when the aortic valve is closed, pressure will decay exponentially with a characteristic decay time RC (see below). Frank’s goal was to derive cardiac output. With the characteristic decay time RC, derived from the aortic pressure in diastole and an independent estimate of total arterial compliance the peripheral resistance could be calculated. Mean flow (i.e. cardiac output) is then simply mean aortic pressure divided by peripheral resistance. Frank estimated total arterial compliance from pulse wave velocity in the aorta. This example shows that Windkessel models and wave transmission of pressure in the aorta give complementary information.
The Windkessel is a so-called lumped model. In other words this lumped model describes the whole arterial system, in terms of a pressure-flow relation at its entrance, by two parameters that have a physiological meaning. One cannot study phenomena that take place inside the arterial tree such as wave travel and reflections of waves, etc.
The two-element Windkessel model tells us that the load on the heart consists of peripheral resistance and total arterial compliance and that both are important.
2 Improvement of Frank’s Windkessel: the three-element Windkessel
In the 1930s and 1940s a number of researchers tried to improve the two-element Windkessel by adding resistance and/or inertance terms and by adding effects of reflected waves. However, a good physiological basis of these models was lacking [34, 80].
With the development of the electromagnetic flow meter and thus measurement of aortic flow, it became clear that in systole the relation between pressure and flow was poorly predicted by the two-element Windkessel . Wetterer [77, 78, 79, 80] had already noticed, but not quantified, the shortcomings of the two-element Windkessel. In 1956, Wetterer  even showed the quantitative difference between measured and predicted pressure in systole, but he did not suggest what the determinant of this pressure difference is.
The characteristic impedance has the same dimension as a resistor and is therefore often represented as a resistor. However, the characteristic impedance is not a resistance, and can only be interpreted in terms of oscillatory phenomena. This means that the ratio of mean pressure over mean flow (i.e. 0 Hz) which equals R, is in the three-element Windkessel R + Z c. The use of a resistor for characteristic impedance also causes errors in the low frequency range of the input impedance . However, since characteristic impedance is, in the systemic circulation of all mammals, about 5–7% of peripheral resistance  the errors are small.
Wang et al.  recently analyzed aortic pressure and flow in the following way. In diastole, when the two-element and three-element Windkessels behave similarly, (with closed valves characteristic impedance does not play a role) these authors fitted the Windkessel and determined total peripheral resistance and total arterial compliance. When this two-element Windkessel was then applied to the entire heart beat a difference was found in systole between the measured pressure and the pressure derived from the two-element Windkessel. This pressure difference, called excess pressure, and according to the definition of Lighthill  a pressure proportional to velocity (or flow), is indeed similar in shape to the measured velocity. The value of this resistance is close to the characteristic impedance of the aorta. This result is thus proof, in the time domain, that addition of the characteristic impedance to the two-element Windkessel, thereby resulting in the three-element Windkessel, is necessary to describe pressure and flow throughout the entire cardiac cycle. We interpret this result as support for the three-element Windkessel with characteristic impedance of the aorta (see Fig. 4) as the third element. However, Wang et al., interpret this resistor as the resistance of the conduit arteries. By doing so this would mean that all large conduit arteries would be described by the proximal resistance. This proximal resistance and arterial compliance would reside within the same (conduit) vessels and cannot be separated.
Fogliardi et al.  compared the three-element Windkessel with constant compliance and with pressure-dependent compliance. They concluded that “the nonlinear three-element windkessel cannot be preferred over the traditional linear version of this model”. Thus the three-element Windkessel suffices in most studies .
3 The four-element Windkessel
In an attempt to reduce the errors in the low frequency range, introduced by the characteristic impedance, a fourth element of the Windkessel has been proposed , an element originally suggested by Burattini and Gnudi . This fourth element (Figs. 4, 5) is an inertance equal to the addition of all inertances in the arterial segments, i.e. total arterial inertance . While aortic characteristic impedance takes into account the compliance and inertance of the very proximal ascending aorta, total arterial compliance and total arterial inertance are the summation of all compliances and inertances in the entire arterial system. The total arterial inertance only affects the mean term and very low frequency behavior of the input impedance, i.e. the frequency range where the three-element Windkessel is inaccurate.
Other investigators have introduced an inertance in series with the characteristic impedance [4, 5, 7, 25, 32, 57]. This series inertance does not affect the arterial input impedance at low frequencies but at high frequencies. In theory this inertance implies an increase in the impedance modulus with frequency in the high frequency range, but in practice the inertance is chosen such that this effect is small.
The four-element Windkessel has been used by Segers [51, 53, 54, 55] and Burattini . In practice it turns out that the inertance is very difficult to estimate which is an argument to prefer the three-element Windkessel.
Burattini [6, 9] tested several other lumped models for the peripheral arterial system: One lumped model consists of a peripheral resistance with, in parallel, a series resistor and compliance. This arrangement leads to the same form of input impedance, as the three-element Windkessel, namely (a + jωb)/(1 + jωc), with a, b, and c constants. The advantage of this particular model is that at 0 Hz, the peripheral resistance is correctly modeled. The resistance in series with compliance is based on viscoelastic arterial properties. Thus the elements have a different meaning than those of the three-element Windkessel model and this model is therefore not an improvement of Frank’s two-element Windkessel model.
Comparison of Windkessel models
Poor at high freq.
Small error at low freq.
Good at all freq.
Good (PP method)
Difficult to estimate
Resistance + charact imp
We conclude that the three-element Windkessel is a necessary improvement of the two-element Windkessel and can model the global aspects of the arterial system with physiologically based parameters.
4 Use and clinical relevance of the Windkessel
The Windkessel teaches us that the main parameters describing the arterial system are peripheral resistance, total arterial compliance and aortic characteristic impedance. In 1997 [3, 36], the importance of pulse pressure in the prediction of cardiovascular mortality and morbidity was shown and the important role of compliance was recognized. It is therefore implicit that, for instance in hypertension, these three Windkessel parameters play a role. Total arterial compliance  and, to a lesser extent characteristic impedance, are now important parameters under investigation.
Ambulatory arterial stiffness index. The Windkessel can be used to clarify the meaning of parameters used in epidemiological studies. An example is the ambulatory arterial stiffness index . Assuming the Windkessel predicted decay time of aortic pressure in diastole, as RC, Westerhof et al.  derived this stiffness index  from first principles and showed that, despite its name, it should be regarded as a ventriculo-arterial coupling factor rather than a compliance estimate.
The Windkessel used to derive cardiac output. Wesseling et al.  developed a method to derive cardiac output from pressure using the three-element Windkessel. The characteristic impedance and total arterial compliance are given as non-linear functions of pressure based on the data of Langewouters [26, 27], using gender, age, body length and weight. An initial value for peripheral resistance in assumed. Stroke volume, SV, is then calculated from pressure. Cardiac output is set to SV·HR and peripheral resistance is calculated as mean pressure over this cardiac output estimate. This resistance value is inserted again in the three-element Windkessel and SV for the next beat is calculated. It turns out that after a few beats convergence is obtained and the true resistance is found: mean pressure over this resistance gives cardiac output. The non-linear behavior of the compliance and characteristic impedance combined with the adaptive peripheral resistance assure excellent tracking of cardiac output .
The Windkessel used as load in artificial heart and valve studies. The (nonlinear) Windkessel, has also been used in studies on ventricular assist devices and setups to test artificial valves [1, 12, 14, 19, 20, 21, 38, 48, 49]. However, in this field many other models of the arterial system are used as well.
Sensitivity to independent arterial and cardiac parameters
Total arterial compliance
The Windkessel used as peripheral bed model. The three-element Windkessel has been used as a simplified representation of peripheral beds in distributed models . However, it should be kept in mind that the characteristic impedance is only a real quantity for large vessels, making the three-element Windkessel a very rough representation of a peripheral bed. Windkessel models with an inertance term have also been used a terminal impedance of the aorta [6, 9].
The Windkessel used in end-ejection identification. Using aortic pressure as input, an uncalibrated flow curve can be calculated from which the time of end-ejection can be identified unambiguously. Since only an uncalibrated flow is required, the parameter estimation reduces to the determination of two time constants which can be conducted in real time .
The windkessel used in solving the outflow problem in impedance cardiography. In impedance cardiography, Stroke volume is estimated from the changes in electrical impedance of the thorax [24, 44]. Since blood both enters and leaves the thorax simultaneously, impedance changes cannot be solely attributed to inflow. To solve this so-called outflow problem , a Windkessel model was used .
The Windkessel used in assessing right ventricular afterload. Windkessel models have also been used extensively to describe the pulmonary vascular bed . Although most of this research on the pulmonary circulation is preclinical, Lankhaar et al.  have shown that the Windkessel model can be used to clinically assess differences between groups of patients with different forms of pulmonary hypertension. However, the windkessel turned out to be too simple a model to be able to classify individual patients.
5 Estimation of total arterial compliance
Several methods, based on the Windkessel have been proposed to estimate total arterial compliance . These methods are:
The Stroke volume over pulse pressure method. This method is rather old [45, 47] but has been reintroduced recently . If the vascular periphery could be completely blocked, i.e. resistance infinite, Stroke volume, ΔV, would increase pressure by ΔP and the ratio would give total arterial compliance: C = ΔV/ΔP. Since part of the Stroke volume leaves the arterial system through the periphery, this ratio overestimates compliance. The overestimation may be as large as 60% [50, 56].
The pulse pressure method [59, 62] is based on fitting the systolic and diastolic pressures, as predicted by the two-element Windkessel with measured aortic flow as input, to the measured values of systolic and diastolic pressure. Although the two-element Windkessel does not produce correct wave shapes, its low frequency impedance is close to the actual impedance, while in three-element Windkessel (by assuming the characteristic impedance as a resistor) introduces errors at the low frequencies. The systolic and diastolic pressure are mainly determined by low frequencies and thus predicted accurately by the two-element Windkessel. The pulse pressure method has been compared with the area method and the SV/PP method and found to be the superior one .
The parameter estimation method fits the three-element or four-element Windkessel to measured pressure and flow as a function of time. When aortic flow is fed into the Windkessel model the pressure is predicted. This pressure can be compared to the measured pressure. By minimization of the difference between predicted and measured pressure (i.e. the sum of squared errors), the best Windkessel parameters are obtained. In this way, all the Windkessel parameters can be derived including a good estimate of characteristic impedance. Using the three-element Windkessel the value of compliance is overestimated  but this is not the case using the four-element Windkessel , however, the inertance estimate is usually poor. Also the inverse procedure may be followed, pressure can be fed into the Windkessel model and optimization of flow is then performed .
The input impedance method is similar to the previous method, but is carried out in the frequency domain. The input impedance of the three-element or four-element Windkessel model is fitted to the measured input impedance.
The transient method  can be applied when pressure and flow are not in the steady state. In a non steady state peripheral resistance cannot be calculated from mean pressure and mean flow, because aortic flow is not equal to peripheral flow. Using the three-element Windkessel with flow as input, pressure may be calculated while storage of blood in the large conduit arteries is accounted for. By curve fitting of the Windkessel parameters to obtain minimal difference between measured and predicted pressure the Windkessel parameters can be estimated accurately.
General remarks. In the methods where the RC-time is derived, the resistance should be calculated from mean pressure and flow, and compliance is then RC-time divided by R.
It should be emphasized that all Windkessel-based methods rely on accurate pressure measurement in the proximal aorta. The first three methods require accurate pressure measurement and mean flow (cardiac output), while the other methods require ascending aortic pressure and flow wave shape.
Quick et al. have circumvented the use of a lumped model as assumed in all methods described above by incorporating transmission of the pressure wave, as in the real arterial tree, to estimate arterial compliance . Recently, they also showed that for wavelengths longer than the arterial tree, distributed models will be reduced to the lumped Windkessel . Such long wavelengths or high wave speed implies that all pressures and flows behave similarly.
6 Windkessel as full description of the arterial system
Arterial input impedance gives a comprehensive description of the arterial system as load on the heart. However, impedance with its modulus and phase, and its frequency dependence is difficult to interpret. Using the three-element Windkessel a comprehensive description of the arterial load is obtained as well. The three elements are derived and have a physiological meaning and one can see which part of the arterial tree is changed. For instance, a decrease in arterial compliance is not easily seen in the impedance and not easily expressed quantitatively either. Derivation of the Windkessel immediately gives quantitative information on total arterial compliance.
The integrated description of the entire arterial system by means of a lumped system like the Windkessel, can adequately describe pressure-flow relations at the entrance of the system, but pressures within these models have little meaning. e.g., the measurement of pressure distal of the characteristic impedance for instance, does not represent the pressure in the more distal vascular system.
The Windkessel is being used in the so-called Physiome, which attempts to develop mechanistic biophysical models within a unifying framework obedient to fundamental mechanical and physicochemical principles . See, for example, the National Institutes of Health “Roadmap for Medical Research in the 21st Century” (http://nihroadmap.nih.gov) mechanistic systems approach to biological sciences .
The windkessel helps to interpret impedance. We interpret the input impedance of the arterial system as follows. For mean pressure and flow and very low frequencies, frequencies <0.2 HR, the distal periphery, i.e. peripheral resistance determines the input impedance. For intermediate frequencies, 0.2 HR < frequencies <3 HR, the more proximal part of the arterial system begins to determine impedance, i.e. total arterial compliance is the main determinant. For very high frequencies, frequencies >3 HR, only the very proximal part of the arterial system, i.e. the aorta contributes, in terms of its characteristic impedance. The characteristic impedance of the aorta is the impedance of the aorta when no reflections exist, apparently at high frequencies the reflections from the many reflection sites return at random times and cancel. Thus the higher the frequency the closer you ‹look’ into the arterial system. The three-element Windkessel indeed contains these three elements.
7 Limitation of the Windkessels
The Windkessel is a lumped model of the arterial system or part thereof. Wave transmission and wave travel cannot be studied. Blood flow distribution and changes in the distribution cannot be represented. Effects of local vascular changes, e.g., change in aortic compliance while other vessels are not affected, cannot be studied. The measurement of pressure distal of the characteristic impedance does not represent the pressure in the more distal vascular system.
We have discussed the main aspects and use of the Windkessel. The three-element Windkessel can adequately describe the pressure-flow relations at the entrance of the arterial system. It is a lumped model and has a limited number of physiologically meaningful parameters. These parameters offer better insight into arterial function than input impedance. In contrast to distributed models [42, 64, 71] Windkessel models are easier to construct as hydraulic load on isolated hearts or assist devices. The Windkessel is also easier to use than distributed models when ventriculo-arterial coupling is studied. The Windkessel can be used for the systemic arterial system and the pulmonary arterial bed of all mammals.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.
- 3.Benetos A, Safar M, Rudnichi A et al (1997) Pulse pressure: a predictor of long-term cardiovascular mortality in a French male population. Hypertension 30:1410–1415Google Scholar
- 4.Broemser P (1932) Beitrag zum Winkessel Theory des Kreislaufs. Z Biol 93:149–163Google Scholar
- 8.Burattini R, Knowlen GG, Campbell KB (1991) Two arterial effective reflecting sites may appear as one to the heart. Circ Res 68:85–99Google Scholar
- 10.Burkhoff D, Sagawa K (1986) Ventricular efficiency predicted by an analytical model. Am J Physiol 250:R1021–R1027Google Scholar
- 11.Chemla D, Hebert JL, Coirault C et al (1998) Total arterial compliance estimated by Stroke volume-to-aortic pulse pressure ratio in humans. Am J Physiol 274:H500–H505Google Scholar
- 15.Elzinga G, Westerhof N (1973) Pressure and flow generated by the left ventricle against different impedances. Circ Res 32:178–186Google Scholar
- 17.Fogliardi R, Di DM, Burattini R (1996) Comparison of linear and nonlinear formulations of the three-element Windkessel model. Am J Physiol 271:H2661–H2668Google Scholar
- 18.Frank O (1899) Die Grundform des arteriellen Pulses. Z Biol 37:483–526Google Scholar
- 20.Glower JS, Cheng RC, Giridharan GA et al (2004) In vitro evaluation of control strategies for an artificial vasculature device. Conf Proc IEEE Eng Med Biol Soc 5:3773–3776Google Scholar
- 23.Hoetink A (2005) On the origin of the electrical impedance cardiogram. VU University, AmsterdamGoogle Scholar
- 30.Lighthill MJ (1978) Waves in fluids. Cambridge [Eng.]. Cambridge University Press, New YorkGoogle Scholar
- 31.Liu Z, Brin KP, Yin FC (1986) Estimation of total arterial compliance: an improved method and evaluation of current methods. Am J Physiol 251:H588–H600Google Scholar
- 32.Livnat A, Yamashiro SM (1981) Optimal control evaluation of left ventricular systolic dynamics. Am J Physiol 240:R370–R383Google Scholar
- 33.Lucas CL (1984) Fluid mechanics of the pulmonary circulation. Crit Rev Biomed Eng 10:317–393Google Scholar
- 34.Milnor WR (1989) Hemodynamics. Williams & Wilkins, BaltimoreGoogle Scholar
- 36.Mitchell GF, Moye LA, Braunwald E et al (1997) Sphygmomanometrically determined pulse pressure is a powerful independent predictor of recurrent events after myocardial infarction in patients with impaired left ventricular function SAVE investigators. Survival and ventricular enlargement. Circulation 96:4254–4260Google Scholar
- 39.Murgo JP, Westerhof N, Giolma JP et al (1980) Aortic input impedance in normal man: relationship to pressure wave forms. Circulation 62:105–116Google Scholar
- 40.Nichols WW, O’Rourke MF (2005) McDonald’s blood flow in arteries theoretic, experimental, and clinical principles. Hodder Arnold, London, p 148 ( Distributed in the U.S.A. by Oxford University Press)Google Scholar
- 41.O’Rourke MF (1967) Pressure and flow waves in systemic arteries and the anatomical design of the arterial system. J Appl Physiol 23:139–149Google Scholar
- 42.O’Rourke MF, Avolio AP (1980) Pulsatile flow and pressure in human systemic arteries. Studies in man and in a multibranched model of the human systemic arterial tree. Circ Res 46:363–372Google Scholar
- 43.Quick CM, Berger DS, Noordergraaf A (1998) Apparent arterial compliance. Am J Physiol 274:H1393–H1403Google Scholar
- 45.Randall OS, Esler MD, Calfee RV et al (1976) Arterial compliance in hypertension. Aust N Z J Med 6(suppl 2):49–59Google Scholar
- 47.Rosen IT, White HL (1926) The relation of pulse pressure to Stroke volume. Am J Physiol 78:168–184Google Scholar
- 48.Rutten MCM, Wijlaars MW, Strijkers GJ et al. (2002) The valve exerciser: a mechanics-based bioreactor for physiological loading of tissue-engineered aortic valves. In: Abstracts of the 13th conference of the European society of biomechanics, Wroclaw, Poland, 2002Google Scholar
- 50.Segers P, Brimioulle S, Stergiopulos N et al (1999) Pulmonary arterial compliance in dogs and pigs: the three-element windkessel model revisited. Am J Physiol 277:H725–H731Google Scholar
- 52.Segers P, Morimont P, Kolh P et al (2002) Arterial elastance and heart-arterial coupling in aortic regurgitation are determined by aortic leak severity. Am Heart J 144:568–576Google Scholar
- 54.Segers P, Stergiopulos N, Westerhof N (2000) Quantification of the contribution of cardiac and arterial remodeling to hypertension. Hypertension 36:760765Google Scholar
- 55.Segers P, Stergiopulos N, Westerhof N (2002) Relation of effective arterial elastance to arterial system properties. Am J Physiol Heart Circ Physiol 282:H1041–H1046Google Scholar
- 57.Sharp MK, Pantalos GM, Minich L et al (2000) Aortic input impedance in infants and children. J Appl Physiol 88:2227–2239Google Scholar
- 58.Simon AC, Safar ME, Levenson JA et al (1979) An evaluation of large arteries compliance in man. Am J Physiol 237:H550–H554Google Scholar
- 60.Stergiopulos N, Meister JJ, Westerhof N (1995) Evaluation of methods for estimation of total arterial compliance. Am J Physiol 268:H1540–H1548Google Scholar
- 61.Stergiopulos N, Meister JJ, Westerhof N (1996) Determinants of Stroke volume and systolic and diastolic aortic pressure. Am J Physiol 270:H2050–H2059Google Scholar
- 62.Stergiopulos N, Segers P, Westerhof N (1999) Use of pulse pressure method for estimating total arterial compliance in vivo. Am J Physiol 276:H424–H428Google Scholar
- 63.Stergiopulos N, Westerhof BE, Westerhof N (1999) Total arterial inertance as the fourth element of the windkessel model. Am J Physiol 276:H81–H88Google Scholar
- 65.Suga H, Sagawa K, Shoukas AA (1973) Load independence of the instantaneous pressure–volume ratio of the canine left ventricle and effects of epinephrine and heart rate on the ratio. Circ Res 32:314–322Google Scholar
- 66.Toorop GP, Westerhof N, Elzinga G (1987) Beat-to-beat estimation of peripheral resistance and arterial compliance during pressure transients. Am J Physiol 252:H1275–H1283Google Scholar
- 68.Wang JJ, O’Brien AB, Shrive NG et al (2003) Time-domain representation of ventricular-arterial coupling as a windkessel and wave system. Am J Physiol Heart Circ Physiol 284:H1358–H1368Google Scholar
- 69.Wesseling KH, Jansen JR, Settels JJ et al (1993) Computation of aortic flow from pressure in humans using a nonlinear, three-element model. J Appl Physiol 74:2566–2573Google Scholar
- 72.Westerhof N, Elzinga G (1991) Normalized input impedance and arterial decay time over heart period are independent of animal size. Am J Physiol 261:R126–R133Google Scholar
- 73.Westerhof N, Elzinga G, Sipkema P (1971) An artificial arterial system for pumping hearts. J Appl Physiol 31:776–781Google Scholar
- 75.Westerhof N, Murgo JP, Sipkema P et al (1979) Quantitative cardiovascular studies. In: Hwang NHC, Gross DR, Patel DJ (eds) Quantitative cardiovascular studies. University Park Press, BaltimoreGoogle Scholar
- 76.Westerhof N, Stergiopulos N, Noble MIM (2005) Snapshots of hemodynamics an aid for clinical research and graduate education. Springer, New YorkGoogle Scholar
- 77.Wetterer E (1940) Quantitative Beziehungen zwischen stromstärke und Druck im natürlichen Kreislauf bei zeitlich variabler Elasticität des Artriellen Windkessels. Z Biol 100:260–317Google Scholar
- 78.Wetterer E (1954) Flow and pressure in the arterial system, their hemodynamic relationship, and the principles of their measurement. Minn Med 37:77–86Google Scholar
- 79.Wetterer E (1956) Die Wirkung der Herztätigkeit auf die Dynamik des Arteriensystems. Verh Dtsch Ges Kreislaufforsch 22:26–60Google Scholar
- 80.Wetterer E, Kenner T (1968) Grundlagen der Dynamik des Arterienpulses. Springer, BerlinGoogle Scholar