Filename
stringlengths 22
64
| Paragraph
stringlengths 8
5.57k
|
|---|---|
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Four BL Lac objects have been detected in the TeV energy range: Mrk 421 (Punch et al. 1992), Mrk 501 (Quinn et al. 1996), 1E S2344+514 (Catanese et al. 1998) and PKS 2155-304 (Chadwick et al. 1999). Recently, the spectrum of Mrk 501 has been measured up to 24 TeV by the HEGRA telescopes (Konopelko et al. 1999). Several of the EGRET AGN show γ-ray variability with time scales of 1 day (Kniffen et al. 1993) at GeV energies. The TeV γ-ray emission of two BL Lacs shows very rapid 15 minutes has been variability. For Mrk 421, variability on a time scale as short as reported (Gaidos et al. 1996). In the case of Mrk 501, variability on a time scale of a few hours was observed during the 1997 high level of activity, and there is evidence of a 23 day periodicity (Protheroe et al. 1998, Hayashida et al. 1998) interpreted in terms of a binary black hole model for the central engine by Rieger and Mannheim (2000). These variability timescales place important constraints on the models. For example, the synchrotron self-Compton (SSC) model appears to be just consistent with recent multi- wavelength observations of Mrk 421 and Mrk 501 during flaring activity (Bednarek and Protheroe 1997, 1999). However, the allowed range of physical parameters (Doppler factor and magnetic field) is rather small, and this mechanism may well be excluded by future observations. For a recent review of TeV γ-ray astronomy see Kifune (2002). Rapid variability in intra-day variability (IDV) sources is a long-standing problem as it implies apparent brightness temperatures in the radio regime which may exceed 1017 K or relativistic beaming with extremely high Doppler factors, coherent radia- tion mechanisms, or special geometric effects (Wagner and Witzel 1995). The very rapid flaring observed at TeV and X–ray energies during flaring activity in blazars also presents a challenge for any model and suggests a re-examination of mechanisms which may cause very rapid variability would be worthwhile. In this paper I concentrate on how the observed variability time is related to the geometry and motion of the emis- sion region, and thus to the photon energy density in the emission region. The blazar emission mechanisms to be discussed include: a shock excited emission region, bent jets, a shock propagating along a jet containing a helical structure and illuminating parts of the helix by enhanced interactions/emission of radiation such that the emis- sion regions move along helical paths, and highly oblique conical shocks in the jet. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Together with geometry-specific time delays and variable Doppler boosting associated with relativistic motion of the emission region along a curved trajectory, it may well be possible to explain the observed flaring activity and high brightness temperatures. Another possibility briefly discussed in the context of bent jets and conical shocks is that jets may be fueled on an irregular time scale. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | The observed variability time ∆tobs, and some assumed or estimated Doppler factor D is often used to estimate the jet-frame source radius, R′ Dc∆tobs. The jet-frame ≈ L′/4πR′2c, where L′ is photon energy density is then usually assumed to be U ′ phot ≈ the jet-frame bolometric luminosity given by L′ = D−44πd2 LF , F being the observed bolometric flux, and dL being the luminosity distance. However, this approach can lead to large systematic errors in the jet-frame photon energy density. This is impor- tant because the energy density of photons of the low energy part of the SED may determine the energy losses of electrons, and the rate of up-scattering to γ-ray ener- gies in SSC models, and the rate of proton-photon collisions in hadronic models (see, e.g., Mannheim and Biermann 1989, Protheroe 1997, Mannheim et al. 2001, M¨ucke and Protheroe 2001, M¨ucke et al. 2002). If the emission region is optically thin, it may also have consequences for IDV sources as I will show that it is quite possible that the photon energy density responsible for the so-called Compton catastrophe may actually be lower than usually estimated. In the following sections I shall discuss how the variability time is related to the emission region geometry and intrinsic jet-frame variability, show how this can lead to large systematic errors in the jet-frame photon energy density, and discuss other scenarios which can lead to rapid variability and high fluxes. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | I shall discuss first a relativistic jet pointing at angle θ with respect to the line of sight to the observer, in which the emission region is a cylinder of radius r and jet-frame length ℓ′ moving along the jet with the jet’s Lorentz factor Γ = 1/√1 β2. To work out how the observed lightcurve depends on the emission region geometry and the duration of the emission in the jet frame (primed coordinates) we first consider in the observer frame (unprimed coordinates) the events corresponding to: (i) the emission zi = 0) with this event from centre of cylinder at (t ≡ defining the origin of coordinates in both frames, i.e. (t′ i = 0, y′ x′ ≡ y′ z′ i = 0, z′ i = 0) ; (ii) the emission from some arbitrary point in the cylinder at a later time (tii, xii, yii, zii) ; (iii) the arrival at the telescope, located in the x–y plane, of the photon emitted in event (i); (iv) the arrival at the telescope of the photon emitted in event (ii) (see Fig. 1). |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | What we have learned from this discussion is that multiplying c∆tobs by the Doppler factor might give the jet-frame intrinsic variability time or one of the dimensions of the emission region (possibly multiplied by some unknown factor). One is tempted to ask phot = L′/4πR′2c in order to estimate if R′ the jet-frame photon energy density. We shall discuss this point further in Section 4. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | The Monte Carlo method allows the accurate calculation of expected light curves for any emission region geometry and intrinsic source variability. The emission region is modeled in the jet frame, and is represented by N “particles”, each of which emits pre- cisely one “photon”. The number density of the particles models the geometry of the emission region, and each particle emits its photon at a time determined by the emis- sion region geometry and variability model. The jet-frame 4-position of each photon emission event (t′ i), i = 1, 2, . . . N, is determined by model. The emission re- gion moves in the x-direction with Lorentz factor Γ, and the observer-frame 4-position of each photon emission event (ti, xi, yi, zi), i = 1, 2, . . . N, is obtained by Lorentz trans- formation. The arrival time of each photon is calculated for a given viewing angle θ using Eq. 1, and is binned to give the lightcurve. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Two emission region geometries are considered: a solid cylinder, and a spheroidal 3D gaussian. We also consider four intrinsic jet-frame time distributions: a simulta- neous burst (violates causality), a simultaneous gaussian pulse (violates causality) and simultaneous emission at plane shock. I shall investigate effects of varying the Doppler factor D, viewing angle θ, jet Lorentz factor Γ, and shock speed βshock. Finally, the effect of smoothing due to acceleration/radiation time delays is discussed. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | If the emission region is “solid”, i.e. the emissivity is constant inside the emission region and zero outside, the lightcurve may be peaked and have finite duration to reflect the sharp edges of the emission region, and the shape of the lightcurve will change with with viewing angle. The Monte Carlo results for the cylinder, when plotted such that the time of observation is divided by the expected variability time ∆tobs given by Eqs. 8 and 10, is shown in Fig. 2(a) for a simultaneous burst (∆t′ = 0) , and in Fig. 2(b) for shock excitation. As can be seen, Eqs. 8 and 10 are verified by the Monte Carlo results. In both cases the shape of the lightcurve depends strongly on the viewing angle. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 2: Lightcurve from cylindrical emission region with ℓ/2r = 0.1 moving with Lorentz factor Γ = 10 as observed at various viewing angles: (a) Simultaneous burst for θ′ = 0◦ (solid), 36◦ (dotted), 72◦ (short-dashed), 108◦ (short-dashed, identical to 72◦), 144◦ (dotted, identical to 36◦), θ′ = 360◦ (long-dash, identical to 0◦); (b) Excitation by a plane shock moving at speed β′ shockc = 0.5c for θ′ = 0◦ (solid), 36◦ (dotted), 72◦ (short-dash), 108◦ (dot-dash), 144◦ (dot-dot-dot-dash), θ′ = 360◦ (long-dash, identical to 0◦). |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | with σx = σr corresponding to a spherical gaussian distribution, σx < σr to an oblate spheroidal gaussian distribution, and σx > σr to an prolate spheroidal gaussian dis- tribution. For the case of a simultaneous gaussian pulse, the probability of emission at jet-frame time t′ to (t′ + dt′) is p(t′)dt′, where p(t′) = exp( 2)/√2πσt is independent of position. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Note the similarity to Eq. 8 except that the terms are added in quadrature as the standard deviation is required instead of the maximum duration of the pulse which is, theoretically, infinite. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 3: Lightcurve due to excitation by a plane shock of a spherical density moving with Lorentz factor Γ = 10 for β′ shock = 0.5 and various viewing angles shown is seen to be a normal distribution. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | can be used for this purpose, where t′ del represents the delay between shock passage and emission by the radiating particles, ∆t′ gain represents the the duration of the acceleration following shock passage. In the case of a pre-existing population of thermal electrons shock heating is essentially instantaneous (∆t′ loss represents the time- scale for cooling by thermal bremsstrahlung radiation. Similarly, for a pre-existing population of relativistic electrons, the passage of the shock will essentially instantly gain = 0) and t′ increase the magnetic field (∆t′ loss would represent the time-scale for energy losses by synchrotron emission. If particle acceleration is required then ∆t′ gain > 0. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | The solid curves, which agree with the Monte Carlo results (histograms), are obtained from Eqn. 18. As we see, if the time-scales for the acceleration/radiation process (∆t′ gain and t′ loss) are much less than the time-scale σ′ associated with the shock passage and dimensions of the emission region then the light-curve will be symmetrical, and in the case of a gausian spheroidal emission region will be a gaussian distribution with stan- dard deviation σ (leftmost histogram). If the time-scales for the acceleration/radiation process are much larger than σ′, then the light-curve will reflect that of the accel- eration/radiation process, i.e. Eqn. 16 (rightmost histogram). The middle histogram shows an intermediate case. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 4: Excitation of an oblate spheroidal gaussian density (σx/σr = 0.2) moving with Lorentz factor Γ = 10 by a plane shock with jet-frame speed β′ shockc = 0.3c, followed by particle acceleration and radiation represented by Eq. 16 with ∆t′ gain = t′ loss for t′ loss = 0.1σr/c, σr/c and 10σr/c (as indicated) for viewing angle θ = 5◦. Solid curves give the result of convolution (Eqn. 18). |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 5: Lightcurve of 3C 454.3 at 37 GHz (Salonen et al. 1987, Terasranta et al. 1992). The fit (solid curve) comprises a background level (horizontal dotted line) and several flare components (dotted curves) with individual flare lightcurves given by Eq. 16 with ∆tgain = tloss = 0.75 yr. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 6: X-ray lightcurve of 1ES1959+65 (Giebels et al. 2002). The fit (solid curve) comprises a background level (horizontal dotted line) and several flare components (dotted curves) with individual gaussian lightcurves with standard deviations σ = 2.5 d. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | brightness temperatures well in excess of the limit TB < 1012 K imposed by Compton scattering (Kellermann and Pauliny-Toth 1969, Kardashev 2000) when the photon energy density in the emission region reaches the energy density in the magnetic field. One extreme example is PKS 0405-385 (Kedziora-Chudczer et al. 1997) which has 1014 K after correcting for interstellar scintillation (see also Walker 1998) TB > 5 requiring a Doppler factor of D = 103 to satisfy the brightness temperature limit. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | To illustrate how critically the energy density depends on the geometry, I shall consider the case of the jet-frame emissivity following a spheroidal gaussian density. Provided the emission is optically thin, as is almost certainly true for the optical–X-ray synchrotron hump in the SED of blazars, then it is straightforward to calculate the average energy density from the emission region geometry and the luminosity. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | The result giving the jet-frame average photon energy density versus σx is shown as the thick solid curve in Fig. 7 for the case of a gaussian spheroidal density. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 7: Jet-frame average photon energy density versus σx. Thick solid curve shows actual density calculated using the Monte Carlo method. “Observed” density is shown by the shaded areas for σt = 0 (top shaded area), σr, 4σr, 8σr, and 16σr (bottom shaded area). See text for further details. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Note that σ (given by Eq. 14) depends on θ′, σt and σx so that the “observed” jet- frame average photon energy density depends also on θ′, σt and σx as well as D. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | U ′ h Although the radio emission in IDV sources is usually assumed to be optically thick, if this is not the case then the above result may also have implications for IDV sources as the photon energy density responsible for causing the brightness temperature limit may actually be a few orders of magnitude lower than estimated on the basis of time- variability, and in that case much lower Doppler factors would be required to avoid the Compton catastrophe. I explore this further in a separate paper (Protheroe 2002). |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 8: Doppler factor raised to 4th power representing bolometric lightcurve due to emission region moving along a bent jet. The jet Lorentz factor is Γ = 10, the bent jet is approximated by part of a circle around which jet plasma moves with angular 0.3◦ (solid curve), η = 27◦ velocity ω, and the emission region is observed at angle η (chain curve), η = 81◦ (dashed curve). |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | VLBI observations show that helical jets or helical structures in jets may be fairly com- mon in AGN (Rantakyro et al. 1998), and theoretical studies have shown that wave-like helical structures can occur as a result of jet precession (Hardee 2000). Several papers discuss helical jet models or the application of helical models to specific sources (Ca- menzind 1986, Rosen 1990, Tateyama et al. 1998, Qian et al. 1992, Schramm et al. 1993, Steffen et al. 1995, Villata and Raiteri 1999). Certainly helical jets or struc- tures would be important in determining the lightcurve of γ-ray and neutrino emission from blazars, and various suggestions have been made about the mechanisms involved (Despringre and Fraix-Burnet 1997, Marcowith et al. 1995). |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | I consider a filamentary helical structure embedded in the jet with Lorentz factor Γ whose axis coincides with the jet axis and is excited by a plane shock travelling along the jet with jet-frame speed β′ shockc. The helical structure could be, for example, a flux tube containing a relatively high magnetic field, or a tube of high plasma density arising from a density perturbation in the plasma entering the jet. Helical magnetic fields with an Archemedian spiral topology similar to the “Parker spiral” field of the heliosphere may well be expected in AGN jets. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | provided the model parameters give cos φ in the range 1 (cos φ depends on the helix geometry, shock speed, jet Lorentz factor and viewing angle). If cos φ = 1 ± 1 < cos φ < 1 multiple cusps occur, one cusp per helix wavelength will occur, and if otherwise no cusps are present in the light curve. However, if cos φ is close to 1 the lightcurve will be peaked. This is illustrated in Fig. 9 for Γ = 10, β′ shock = 0.5, and λ = r. Fig. 9(a) shows t′ plotted against tobs for five viewing angles, and Fig. 9(b)–(f) shows the resulting lightcurve for each of the five viewing angles (the corresponding Doppler factors are also given). |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Apart from the periodicity, the lightcurves shown in Fig. 9 are reminiscent of those of blazars. All of these lightcurves correspond to instantaneous emission from the point on the helical filament at the time of excitation. The cusps and peaks would in reality be smoothed to some extent by the finite width of any helical structure as well as by any delays associated with acceleration and radiation time scales. Furthermore, as the shock weakens while propagating down the jet the successive peaks/cusps in the lightcurve would decrease in height. One possible scenario could be that there is a succession of shocks at random intervals propagate down the jet, and because individual shocks weaken as they propagate, each shock would cause only one or two peaks (due to only one or two cycles of the helix). In this case, it may not be possible to distinguish between a simple bent structure in the jet and a helical structure – if appropriately aligned, a simple bent structure would cause a cusp in the lightcurve in exactly the same way as a helical structure. An alternative to a helical filamentary structure within the jet is a helical jet which would produce a qualitatively-similar lightcurve, but would have the Doppler factor varying with position along the jet as the viewing angle relative to the local jet direction changes. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Lind and Bladford (1985) have considered the possibility that hotspots seen in VLBI images of radio jets may actually be relativistically moving conical shocks. They ap- plied the relativistic shock jump conditions to a plane parallel flow entering a forward conical shock with cone angle η to find the angle ζ < η with at which the flow initially diverges with respect to the jet axis. Defining the downstream region between the cones with angles ζ and η to be the emission region, and taking account of Doppler boosting, they model the brightness distribution to simulate VLBI images. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | The generation of conical shock structures is often seen to occur in simulations of relativistic jets after the introduction of a fast perturbation (Bowman 1994, Gomez et al. 1997). These structures, which are typically an alternating sequence of forward and reverse conical shocks, are stationary or slowly-moving in the galaxy frame, and have cone angles η 1/Γ. Salvati et al. (1998) discuss emission by a conical shock. They consider the case of a density perturbation, confined to thin flat disk, travelling relativistically along the jet causing particle acceleration and emission where the disk cuts a stationary forward conical shock such as that illustrated in Fig. 10. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | I wish to extend the work of Salvati et al. (1998) by including the reverse shocks, and ultimately a sequence of stationary reverse and forward shocks. In Fig. 11(b) I show the lightcurve for a reverse conical shock having identical parameters as the forward shock already discussed. Because the divergence of the downstream flow is rather small, the lightcurve of the reverse shock is approximately just the lightcurve of the forward shock reflected about tobs = 0. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | I shall consider next the case of a sequence of stationary reverse and forward conical shocks. Although in real AGN jets it may be possible that the downstream flow is re- accelerated to near the original value (depending on conditions external to the jet, and the jet production mechanism), I shall assume that each successive conical shock causes a reduction in the jet Lorentz factor by Γd/Γu = 0.801784. I shall make an additional approximation that the divergence/convergence of the flow caused by the conical shocks can be neglected, and that the jet flow is always parallel to the jet axis. The lightcurve due to a thin, initially flat, disk travelling relativistically along the jet is then calculated by the Monte Carlo method, in which “particles” are placed over the surfaces of all of the cones. Each cone has the same number of particles, and they would appear uniformly distributed across the cross section of the jet when viewed along the jet axis. A side view showing the location of these particles is given in Fig. 12. An initially flat thin disk is launched along the jet with initial Lorentz factor Γ = 10, and each time a part of the disk crosses one of the shocks the Lorentz factor of that part of the disk drops appropriately, distorting the disk. The time at which each “particle” emits its photon is determined by the time at which the (distorted) disk reaches the “particle”. The resulting lightcurve for various viewing angles is shown in Fig. 13(a) where the emission is boosted using the Doppler factor corresponding to the Lorentz factor of the flow immediately downstream of each shock. If the Lorentz factor of the jet decreases at each shock crossing as assumed here then, because the Doppler factor is also reduced at each shock crossing, only the first few conical shocks would be prominent in the lightcurve resulting from a single jet perturbation. This is shown in Fig 13(b) which gives the lightcurve from Fig 13(a) for θ = 9◦ on a linear scale. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | As can be seen, the lightcurve is quasi-periodic. It would be strictly periodic if the downstream flow velocity were identical to the upstream flow velocity. This might be the case if the re-collimmation of the diverging flow from the conical shocks results in restoration of the flow Lorentz factor to near the upstream value, and in this case the flux from successive cycles would be at about the same level, as observed in Markarian 501 flares. The two flow velocities would also be roughly the same if the initial jet Lorentz factor were much higher than Γ = 10 used in Fig. 13(a) such that the intervals between flares due to a pair of conical shocks was roughly the same. However, in this case the flux of successive flares would diminish as the Doppler factor decreases. Turning this argument around, we may have a method of determining the minimum jet Lorentz factor from observational data, e.g. the 23 day periodicity in the 1997 Markarian 501 data (Protheroe et al. 1998, Hayashida et al. 1998) may be used to put a lower limit on Γ. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | a density perturbation with a mean interval of r/c. I have also used an exponential distribution for the strength of each perturbation. The resulting lightcurve looks, at least qualitatively, as good as any model for the variability of fluxes from blazars. Note that even though there is no strict periodicity resulting from the excitation of a sequence of shocks in this case, each flare episode has two strong peaks due to (for this viewing angle) the apex of the first reverse shock cone and the apex of the first forward shock cone, and in each flare the separation between these two peaks is identical. Fourier analysis would therefore show a strong peak at the frequency corresponding to the time interval between these two peaks in a single flare. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Many factors can influence observed variability time. The connection between ∆tobs, Doppler factor and emission region geometry is non-trivial, and so measuring ∆tobs may give, at best, one of the dimensions of the emission region. Using R′ = c∆tobsD phot = L′/4πR′2c may then lead to over-estimation or under-estimtion of the and U ′ jet-frame photon energy density by orders of magnitude. This is clearly of importance in any AGN model in which the low energy photons produced in the jet are targets for interaction of high energy particles or radiation, such as in SSC models and hadronic blazar models. Although not discussed in detail in the present paper, the escape of γ-rays from the emission region depends on the optical depth to photon-photon pair production interactions. This optical depth can be uncertain by orders of magnitude in the same way as the photon energy density, and will also depend on viewing angle. One must therefore be careful when using the observation of apparently unattenuated gamma-rays, and an observed variability time, to place limits on the Doppler factor. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | The uncertainty in the jet-frame photon energy density discussed in this paper may also have implications for the high brightness temperature/Compton catastrophe problem of IDV sources. In this case, it is the energy density of target photons which limits the brightness temperature through the competition of inverse-Compton scat- tering with synchrotron radiation, and the target photon energy density may actually be lower than estimated if the emission region is non-spherical. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Distinguishing between the excitation of a conical and a helical structure by a plane shock would be almost impossible – note the qualitative similarities between lightcurves depicted in Figs. 9(d) and 13(b). Relativistic jet simulations (Bowman 1994, Gomez et al. 1997) do show the presence of conical shocks, and these shocks appear after a large perturbation (e.g. from 4 to 10 in the simulations of Gomez et al. 1997) in Lorentz factor of the matter entering the jet through the nozzle. This can result in a sequence of quasi-stationary to superluminal reverse and forward conical shocks extending from the nozzle to the perturbation as it moves along the jet (Agudo et al. 2001). For the conical shock model (Salvati et al. 1998) discussed here to work, a subsequent perturbation would need to result in a plane shock, or plane thin perturbation of some kind, which could travel along the jet and excite the pre-existing conical shocks. As far as I am aware, whether or not this could occur has not been demonstrated. A similar uncertainty hangs over whether or not helical jet structures, which may themselves be shocks with a twisted ribbon topology, resulting from a perturbation entering the jet through the nozzle, could subsequently be excited by the passage of a plane shock or perturbation. Nevertheless, in both cases, if the viewing angle is favourable pile- ups in tobs, and hence flares, could occur simply as a result of the motion of the conical or helical patterns which may themselves be sites of enhanced emission. Note that in recent 3D relativistic jet simulations, the introduction of a 1 percent helical velocity perturbation at the nozzle results in a helical pattern propagating along the jet at nearly the beam speed (Aloy et al. 1999), and that the conical shocks resulting from a perturbation in jet Lorentz factor can range from being quasi-stationary to superluminal (Agudo et al. 2001). |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | In conclusion, in models for flaring in AGN in which the emission comes from a localized region (blob) co-moving with the jet, time variability is non-trivial to interpret in terms of emission region geometry and Doppler factor. A further complication is that flaring may arise instead due to curved or helical motion of a blob, even if the emission is constant in the istantaneous rest frame of the blob. In this case, apparent flaring is due to the change in viewing angle, and hence Doppler factor. Similarly, if the viewing angle is favourable, relativistic motion of curved or helical filaments or surfaces can lead to observation of flares. Excitation of curved or helical jet structures by shocks or perturbations can also lead to pile-ups in tobs, and hence large apparent increases in flux. Observations of time variability in AGN is therefore non-trivial to interpret and may lead to large systematic errors in estimated jet-frame photon energy density, Doppler factor and the physical parameters of the emission region. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | I thank Peter Biermann for helpful discussions and Alina Donea for a careful reading of the manuscript. I would like to thank the anonymous referees for their suggestions which have led to improvements to the paper. My research is supported by a grant from the Australian Research Council and a grant from the University of Adelaide. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Marcowith, A., Henri G. & Pelletier G., 1995, MNRAS, 277, 681. Mendoza, S. & Longair, M.S. 2001, MNRAS, 324, 149 von Montigny C. et al., 1995, ApJ, 440, 525 Mukherjee R. et al., 1997, ApJ, 490, 116. Protheroe, R.J. 1997, in Accretion Phenomena and Related Outflows, IAU Colloq. 163. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 9: Shock excited helical jet structure for Γ = 10, β′ shock = 0.5, λ = r. (a) t′ vs. tobs for θ = 0.3◦ (solid curve), 1◦ (dotted curve), 3◦ (short dashed curve), 9◦ (dot-dash curve), and 27◦ (dot-dot-dot-dash curve). Resulting lightcurves for (b) θ = 0.3◦, (c) 1◦, (d) 3◦, (e) 9◦, and (f) 27◦. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 10: Side view of the jet showing the location of a stationary forward conical shock, and showing the downstream flow diverging from the jet axis at angle ζ (not to scale). For a viewing angle θ with respect to the jet axis, the angle between the downstream flow direction and the line of sight varies between θa and θb. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 11: Lightcurve of (a) a stationary forward conical shock and (b) a stationary reverse conical shock locally excited by a thin density perturbation travelling along the jet with Lorentz factor Γ = 10 as viewed at angle θ with respect to the jet axis. The cone angle is η = sin−1(1/Γ) and results are given for θ/η = 0.001 (long dashed curves), 0.1 (dot-dot-dot-dash curves), 0.3 (dot-dash curves), 0.5 (short dashed curves), 0.7 (dotted curves), 0.9 (solid curves). The Doppler factor is calculated using the local downstream flow velocity. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 12: Side view of the jet showing the location of “particles” used in the Monte Carlo Method. The location of the conical shocks is clearly evident. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 13: Lightcurve due to excitation of the sequence of stationary conical conical shocks discussed in the text for an initial jet Lorentz factor Γ = 10 (a) as seen at viewing angles θ = 0◦ (solid curve), 3◦ (dotted curve), 6◦ (short dashes), 9◦ (dot-dash) and 12◦ (dot-dot-dot-dash) with respect to the jet axis, and (b) as seen at viewing angles θ = 9◦ and plotted on a linear scale. |
Processed_Factors_Determining_Variability_Time_in_Active_Gal.txt | Figure 14: Lightcurve due to repeated excitation of the sequence of stationary conical conical shocks discussed in the text for an initial jet Lorentz factor Γ = 10 as seen at viewing angles θ = 9◦. The time interval between perturbations entering the jet is sampled from an exponential distribution with mean r/c, and the strength of the perturbation is also sampled from an exponential distribution. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | Publications of the Astronomical Society of Australia (PASA) c(cid:13) Astronomical Society of Australia 2021; published by Cambridge University Press. doi: 10.1017/pas.2021.xxx. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | Abstract We have used the Australia Telescope Compact Array (ATCA) to search for a number of centimetre wave- length methanol transitions which are predicted to show weak maser emission towards star formation regions. Sensitive, high spatial and spectral resolution observations towards four high-mass star formation regions which show emission in a large number of class II methanol maser transitions did not result in any detections. From these observations we are able to place an upper limit of <∼ 1300K on the brightness temperature of any emission from the 31A+–31A−, 17−2–18−3 E (vt = 1), 124–133 A−, 124–133 A+ and 41A+–41A− transitions of methanol in these sources on angular scales of 2′′. This upper limit is consistent with current models for class II methanol masers in high-mass star formation regions and better constraints than those provided here will likely require observations with next-generation radio telescopes. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | where ν is the frequency of the observations and θ is the angular scale of the emission region. For a source to be detectable at a particular frequency requires ei- ther a high brightness temperature, or emission over a large angular scale. Maser sources generally fall into the first of these categories with strong 6.7 and 12.2 GHz methanol masers having brightness temperatures in ex- cess of 1010K (e.g. Menten et al. 1992). The majority of maser searches are made with single-dish telescopes with moderate (around 1 Jy) sensitivity limits, however, while these are efficient for detecting strong masers, they do not place strong upper limits on moderate or low brightness temperature masers ( <∼ 108 K). Sensi- tive, high resolution observations of some rare class II methanol maser transitions have shown that masers with brightness temperatures of less than 108K are likely present in some sources (Krishnan et al. 2013). So rigorous tests of maser models require, sensitive, high resolution observations in order to detect, or limit mod- erate and low brightness temperature masers. The Aus- tralia Telescope Compact Array (ATCA), is an ideal in- strument for undertaking such observations and in this paper we report the results of a search for a number of methanol maser transitions which are predicted to exhibit weak maser emission under some physical con- ditions. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | In order to test and refine maser models and poten- tially the physical conditions in the observed star for- mation regions, we selected a number of potential class II methanol maser transitions in the frequency range 5– 9 GHz. A search for potential new maser transitions is a compromise between the number of different sources observed and the sensitivity of the observations. For the primary target transitions we made observations of four high-mass star formation regions which are known to exhibit moderately strong emission from a range of different class II methanol maser transitions. For the secondary target transitions we made observations of a single source G 345.01+1.79, as there are more class II methanol maser transitions detected toward that source than any other (Ellingsen et al. 2012). |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | quencies of 7.682232(50) and 7.830864(50) GHz respec- tively (Tsunekawa et al. 1995, the numbers in brack- ets represent the uncertainty in the least significant figures). These two transitions were observed towards four high-mass star formation regions (G 339.88-1.26, G 345.01+1.79, NGC6334F and W48), which all ex- hibit strong class II methanol maser emission in mul- tiple transitions. The pointing centre for the observa- tions of each source and the peak velocity of the class II methanol maser transitions is given in Table 1. For each transition, observations consisted of a series of scans of 5 minutes for each target source spread over a range of hour angles. Each scan on a target source was both preceded and followed by a 1 minute scan on a nearby phase calibrator. The total observing time for each of the four target maser sources is around 45 minutes for each of the primary transitions. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | Observations were also undertaken of secondary tran- sitions 31A+–31A−, 17−2–18−3 E (vt = 1) and 41A+– 41A− which have rest frequencies of 5.0053207(6), 6.181589(153) and 8.341622(4) GHz respectively (Breckenridge & Kukolich 1995; Xu & Lovas 1997). These three transitions were observed only towards the G 345.01+1.79 star formation region with onsource du- rations of 4.2 minutes for each transition. Of the 5 transitions observed, the only one which has been de- tected astronomically is the 31 A+–31 A−, for which Robinson et al. (1974) report weak, broad emission to- wards Sgr B2. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | the FULL 4 1024 POL mode which has 1024 spectral channels across a 4 MHz bandwidth, with all po- larization products recorded. This corresponds to a velocity coverage of approximately 155 km s−1 at a frequency of 7.7 GHz and velocity resolution of approximately 0.18 km s−1. The data were reduced using miriad, applying the standard data reduction procedures for ATCA observations. The primary flux density calibration was with respect to the standard scale calibrator PKS B1934-638 using the in-built within miriad (Reynolds 1994). The absolute flux density calibration is expected to be accurate to better than 10%. PKS B1934-638 was also used for bandpass calibration. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | tinuum images and we then used continuum subtrac- tion to extract a spectral line dataset. We imaged a 10 km s−1 velocity range covering all the class II methanol maser emission in each source at a velocity resolution of 0.2 km s−1(see Table 1). Inspection of the image cubes for each of the sources did not reveal any emission in either of the two primary target transitions in any of the four source observed. The RMS noise level in 0.2 km s−1 spectral channels were in the range 27 – 37 mJy beam−1 for all sources and are listed in Ta- ble 1. Figure 1 shows the continuum images formed from the 124–133 A+ dataset after self-calibration for G 345.01+1.79 and NGC6334F. We do not show the continuum image for W48 because as it is very close to the equator, the ATCA as an East-West synthe- sis array has a very elongated synthesised beam and does not produce good images. Comparison of the im- ages in Figure 1 with previous more sensitive observa- tions of these sources at centimetre wavelengths (e.g. Ellingsen et al. 2005) shows similar structure and in- tensity. This demonstrates that despite both of these transitions lying outside the nominal frequency range of the ATCA receiver system (at the time of the observa- tions) the system performed well and the non-detection of the spectral line emission is robust. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | cluster with an angular extent of less than 0.5 arcsec- onds (Krishnan et al. 2013). |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | Sobolev et al. (1997a) used the methanol maser pump- ing model of Sobolev et al. (1997b) to predict which transitions are likely to show maser emission under a range of different physical conditions. In total 6 different pumping model input parameters were varied and the results for 10 different combinations are listed in table 1 of Sobolev et al. (1997a). The following model param- eters were varied, the gas temperature (Tkin), the gas density (nH), the maser beaming factor (ǫ), the dilution of continuum emission from a background Hii region (WHII ), the dust temperature (Td) and the methanol specific column density NM/∆V . The 124–133 A− and 124–133 A+ transitions were predicted to exhibit weak maser emission (brightness temperatures of 104 – 105K) in 6 of the 10 models, more than any of the other methanol transitions at frequencies < 10 GHz which had not previously been detected. The most recent pub- lished class II methanol maser pumping model is that of Cragg et al. (2005), which is based on the earlier work of Sobolev et al., but incorporates a number of improve- ments and refinements, including new collisional data and more energy levels. The Cragg et al. (2005) model also suggests that the 124–133 A− and 124–133 A+ tran- sitions are good candidates for weak masers, with 6 of the 8 sets of input physical parameters they trialled yielding masers with brightness temperatures > 103K, with three of them having brightness temperatures in excess of 104K. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | Figure 1. Radio continuum emission at 7.8 GHz for G 345.01+1.79 (left) and NGC6334F (right). For both sources contour levels are at 2.5, 5, 10, 20, 40 and 80% of the image peak which is 217 and 776 mJy beam−1 for G 345.01+1.79 and NGC6334F respectively. The RMS noise level as are 1.4 and 4.1 mJy beam−1 for G 345.01+1.79 and NGC6334F respectively. The synthesised beam for these observations was approximately 2.0′′× 1.8′′. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | maser transitions observed in our four target sources with the results in table 2 of Cragg et al. (2005), they best match model 3, as it is the only model that pre- dicts moderately strong 37.7 GHz methanol masers, which all these sources have. This model has Tkin = 30K, Td=175K, nH = 107 cm−3, NM/∆V = 3 × 1011 cm−3s and WHII = 0.002. This model predicts bright- ness temperatures of ∼ 6.3 × 103 K for the 124–133 A− and 124–133 A+ transitions. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | Table 1 shows that our most sensitive observations were of the 124–133 A+ transition at 7.83 GHz for which we obtained RMS noise levels in 0.2 km s−1 spec- tral channels of < 30 mJy beam−1 for three of the four sources (we had less time onsource for W48, so the RMS noise is higher). We can be confident that there is no emission stronger than 0.150 Jy beam−1 (5- σ) in this transition in the observed sources and from equation 1 that places an upper limit of ∼ 1100K on the brightness temperature of any emission on angular scales of 2′′ (the angular resolution of our observations). This is significantly less than the predicted brightness temperature in the Cragg et al. (2005) model, however, maser emission generally occurs on much smaller an- gular scales than the 2′′ resolution of our synthesised beam. Very long baseline interferometry observations of class II methanol masers (e.g. Menten et al. 1992; Minier et al. 2002; Harvey-Smith & Cohen 2006) show that the 6.7 GHz class II methanol masers generally have angular sizes in the range 5–50 milliarcseconds. Re- arranging equation 1 for the 124–133 A+ transition, our 0.15 Jy beam−1 upper limit on the flux of the emission from this transition implies θ2Tb < 4330 K arcsecond2. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | This corresponds to upper limits on the brightness tem- perature of 1.7 × 106K for emission on scales of 50 mil- liarcseconds or 1.7 × 108K for emission on scales of 5 milliarcseconds. These limits are both more than a order of magnitude higher than the predictions of any of the models of either Sobolev et al. (1997a) or Cragg et al. (2005) for this transition. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | The observations of the 124–133 A− transition at 7.62 GHz are approximately 20% less sensitive than those for the 124–133 A+ transition. The lower sensi- tivity is believed to be due to poorer receiver perfor- mance at the frequency of that transition (at the time of our observations). The affect of this is that the limits for the 124–133 A− transition are 20% higher, however, this makes no significant difference to the conclusions we can draw. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | It should be noted that the analysis undertaken here utilises data with a 0.2 km s−1 velocity resolution. This is close to optimal for Galactic maser work, as typical line widths are around 0.5 km s−1. Averaging several spectral channels would yield lower RMS noise levels in spectral line cubes, however, the resulting limits would be misleading, as they would not account for the dilu- tion of narrow line emission. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | in a 0.2 km s−1 spectral channel would require observa- tions about 2 orders of magnitude more sensitive than those we report here. This level of sensitivity could be achieved with very long (of order days) integrations on sensitive telescopes with large collecting areas (e.g. the GBT, JVLA or FAST), however, it more realistically lies within the realm of a future SKA survey of line emission in Galactic high-mass star formation regions. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | We have searched for emission from a number of cen- timetre wavelength methanol transitions which are pre- dicted to show weak maser emission towards high-mass star formation regions under some physical conditions. Our observations did not detect emission from any of the transitions in any of the four sources observed with upper limits of approximately 0.15 Jy. The observations reported here are approximately an order of magnitude more sensitive and an order of magnitude higher an- gular resolution than typical single dish searches for maser transitions and so they do represent a significant improvement on many searches for weak, rare class II methanol masers (e.g. Ellingsen et al. 2003; Cragg et al. 2004). Previous observations have demonstrated that more sensitive observations can lead to the detection of weak maser lines, for example Sobolev et al. (2007) detected weak 23.1 GHz class II methanol masers after earlier unsuccessful searches by Cragg et al. (2004). To significantly improve on the limits placed on the 31A+– 31A−, 17−2–18−3 E (vt = 1), 124–133 A−, 124–133 A+ and 41A+–41A− transitions of methanol by the current observations will require next-generation radio astron- omy facilities. |
Processed_A_sensitive_search_for_predicted_methanol_maser_tr.txt | The Australia Telescope Compact Array is part of the Aus- tralia Telescope which is funded by the Commonwealth of Australia for operation as a National Facility managed by CSIRO. This research has made use of NASA’s Astrophysics Data System Abstract Service. AC was funded by a Univer- sity of Tasmania Dean’s Summer Research scholarship. AMS has been supported by the Ministry of Education and Sci- ence of the Russian Federation within the framework of state work “Organization of Execution of Scientific Research”. This work was also supported by Government of the Russian Federation Act 211, contract no 02.A03.21.0006. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | Abstract—The increased adoption of electric vehicles (EVs) has led to the development of vehicle-to-anything (V2X) technologies, including vehicle-to-home (V2H), vehicle-to-grid (V2G), and en- ergy trading of EVs in the local grid. The EV coordination can provide value to the grid and generate benefits for EVs. However, network constraints and uncertainties in renewable energy and demand pose significant challenges to EV coordination and restrict the realization of these benefits. This paper develops a rolling-horizon optimization problem for V2X value stacking to fully unlock the value of EV coordination, considering power net- work constraints (such as voltage limits) and uncertainties in the energy system. By coordinating EVs to perform V2H, V2G, and energy trading, our approach exploits the most valuable services in real-time. We also analyze the expected extra costs caused by the prediction errors to evaluate the impact of uncertainties on V2X value stacking. We validate our value-stacking model using real data from Australia’s National Electricity Market (NEM), ISO New England (ISO-NE), and New York ISO (NY-ISO) in the US. The results show that V2X value stacking achieves significant benefits to EVs through energy cost reduction. The uncertainty in the load has a higher impact on the value-stacking performance than PV generation, indicating the importance of load prediction. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | Recent studies have explored methods to optimally schedule EV charging and discharging to interact with smart homes, local energy markets, and the grid to maximize the value of EVs. For example, Nakano et al. in [4] studied the aggre- gation of vehicle-to-home (V2H) systems to participate in a regulation market using in-vehicle batteries. Bijan et al. in [5] studied vehicle-to-grid (V2G) and EV energy trading in the local grid by analyzing the impact of V2G on reliability, cost, and emissions considering stochastic renewable energy. Shurrab et al. in [6] proposed a vehicle-to-vehicle (V2V) energy-sharing framework to maximize social welfare while meeting the charging demand. Al-Obaidi et al. in [7] designed an EV scheduling method to support peer-to-peer (P2P) energy trading and ancillary services. Thompson et al. in [8] devel- oped a vehicle-to-anything (V2X) value stream framework, including V2H, vehicle-to-building (V2B), and V2G, to study the economic potential of V2X. However, the multiple-value stream of EVs may be restricted by local network constraints, and the results are not convincing if such constraints are not considered. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | Power network constraints have been considered in more recent studies on EV scheduling and coordination. For ex- ample, Mazumder et al. in [9] studied the optimal EV charging/discharging to provision V2G and support voltage management in a distribution network. Affolabi et al. in [10] proposed a two-level market framework for energy trading among EVs in charging stations taking into account power network constraints. Nevertheless, uncertainties in the energy system, such as stochastic renewable energy generation and consumers’ time-varying load, pose significant challenges to EV scheduling, causing infeasible scheduling and network constraint violation. Even with advanced predictions for re- newable generation and load, prediction errors are inevitable and will affect the performance of EV scheduling and coordi- nation. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | Corresponding author: Hao Wang. This work has been supported in part by the Australian Research Council (ARC) Discovery Early Career Researcher Award (DECRA) under Grant DE230100046. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | an RHO method for EVs in an electricity exchange market to tackle the uncertainty of electricity prices. Nimalsiri et al. in [13] proposed an RHO method to handle the uncertainty of EV arrival and departure time for a coordinated EV charging and discharging problem while maintaining the feeder voltage within safety limits. As illustrated above, prediction errors are inevitable, and prediction errors often have a negative impact on EVs enjoying the benefits from V2X services. The aforementioned studies did not consider how each uncertainty affects the system operation and value stacking benefits. In contrast, our work aims to maximize the value of EVs within power network constraints and quantify the impact of pre- diction errors of renewable energy and demand, highlighting the importance of different uncertainties in the value-stacking problem. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | In this paper, we are motivated to develop a V2X value- stacking optimization problem to maximize the economic ben- efits for residential households with EVs, while maintaining the local grid voltage within limits under prediction errors. We adopt the RHO method to solve EV value-stacking decisions in real-time and evaluate the impact of prediction errors of solar PV generation and consumers’ load. We validate our value- stacking optimization using real-world data from Australia’s National Electricity Market (NEM), ISO New England (ISO- NE), and New York ISO (NY-ISO) in the US. The main contributions of this paper are as follows. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | We develop a V2X value-stacking problem for residential EVs to exploit multiple value streams, including V2H, V2G, and energy trading, considering distribution net- work constraints and accounting for prediction errors. We formulate an RHO problem to solve the optimal value-stacking decisions for EV coordination in real-time based on predicted PV generation and energy demand of households using Long Short-Term Memory (LSTM). • We evaluate the impact of prediction errors of PV gener- ation and demand on the EV value-stacking optimization by analyzing the relationship between the extra cost and prediction errors. Our analysis shows that the uncertainty in household energy demand has a greater impact than the PV generation uncertainty, highlighting the importance of energy demand forecasting. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | We validate our V2X value-stacking method by consid- ering two retail tariffs, e.g., time-of-use (TOU) pricing and two-part tariff (TPT) pricing, for EVs, and using real-world data from three markets, including Australia’s NEM, ISO-NE, and NY-ISO in the US. The results show that, amongst all modeled value streams, V2H contributes most to the cost reduction in NEM and ISO-NE. Energy trading contributes most to the cost reduction in NY-ISO. Value-stacking leads to higher cost reduction under TOU pricing than TPT pricing. Energy trading creates the most marginal value compared to V2G and V2H in all three markets. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | the power network model. Section III formulates the rolling horizon value-stacking optimization problem and LSTM-based prediction model for PV generation and residential energy demand. Section IV presents three baseline optimization prob- lems and performance evaluation metrics. Section V evaluates our proposed value-stacking method using real-world data from NEM, ISO-NE, and NY-ISO, respectively. Section VI concludes our work. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | We consider an energy system with multiple households (as prosumers) having EVs connected through a local distri- bution network. Fig. 1 depicts an illustrative system for EV coordination, where households and EVs sit on a radial power network [14]. The orange bold line represents the power from the main grid, and the blue solid line represents the power flow of V2G and energy trading in the local network. We define I = {1, . . . , I} as the set of nodes in the distribution network. Additionally, we denote U = {1, . . . , N } as the set of prosumers, and N is the total number of prosumers. Among prosumers, there is a group of prosumers connected to node i, denoted as Bi ⊆ U. It is noted that each prosumer must belong to one group and the prosumer groups B1, B2, . . . , BI do not overlap with each other, i.e., B1 ∪B2 ∪· · ·∪BI = U and B1 ∩ B2 ∩ · · · ∩ BI = ∅. We denote the operational horizon as H = {1, . . . , H} with evenly-spaced time slots, and consider H = 24 hours for a day. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | where µ ∈ [0, 1] and η ∈ (0, 1] denote the charging and discharging efficiency of prosumer u’s EV, respectively. The charge power and discharge power are denoted as pu,t EVc and pu,t EVd in t-th time slot, and bEVs is the stored energy in prosumer u’s EV. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | u,t where Ru,t, P Buy, and Gu are the upper-bounds of renewable energy, power prosumer u purchase from the local market, and grid power for prosumer u at t. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | Buy + pu,t where the left-hand side corresponds to the power supply from the grid pu,t renew, the local energy market pu,t V2H. The right-hand side represents the total load demand including the inflexible load P u,t Load and the EV charging demand pu,t EVc. Note that prosumer u can also sell power to other prosumers and the grid, and we will introduce it in Section III-A. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | where pi,t is the active power flow from node i−1 to i in time slot t, pi,t Load is the amount of active power that consumed by both the active inflexible load P i,t Inflex and the electricity usage of prosumers belonging to the set Bi in node i at t. Note that pi,t Load can have a negative value when prosumers sell power to the network from node i at t. The maximum and minimum allowed active power are P max DN . The reactive power flow is denoted by qi,t from node i − 1 to i at t, and Qi,t Load is reactive load at node i in t. The voltage of node i at t is denoted as vi,t, V 0,t is the voltage of node 0 at t, ri and xi is the resistance and reactance of branch connected to node i − 1 and i respectively, and V i min are the maximum and minimum value of voltage in each node i. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | be recharged from the main grid and PV. We develop a predict- and-optimize method. In each time slot t, we use an LSTM model to predict the household load and PV generation during the time slot t+1 to H for prosumer u. Using the predicted data, we develop an RHO problem to solve the optimal value- stacking decision for prosumer u over the operational horizon [t + 1, H]. In each time slot t, Prosumer u will implement the real-time decision in t and repeat this process in a rolling manner. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | u,t V2H denotes the maximum allowed V2H power at t. 2) Value Stream of V2G: V2G is a technique to utilize EV batteries to sell energy back to the grid and provide ancillary services, such as spinning reverse [19]. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | Residential load and solar PV generation forecasting will be needed in the value-stacking RHO optimization, presented in Section III-C. The forecasting would affect the scheduling for prosumers. Long Short-Term Memory (LSTM) is a type of recurrent neural network, which has been shown to be promising in time series forecasting [21]. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | For residential load and PV generation time series forecast- ing, we utilize the LSTM network model to predict hourly load and PV generation, respectively. The architecture of the LSTM model is presented in Fig. 3, where it starts with pre- processing for the input data, e.g., historical residential load and PV generation data. We denote X = {x1, x2, x3, · · ·, xT } as the input sequence for the LSTM model, where xt ∈ Rn is a n-dimensional vector of real values at the t-th time slot. To prevent training from diverging, we normalize the training sequence to have zero mean and unit variance. Additionally, for training a straightforward one-layer LSTM network, one needs to set the hyper-parameter of the hidden output dimension n. The hidden output ht at a specific time slot, in this work, is an n-dimensional vector. The sigmoid function is deployed in the fully connected layer to limit the output range from 0 to 1. After implementing denormalization, we can obtain the forecast results for load and PV generation. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | Fig. 3. Architecture of the LSTM model for predicting residential load and PV generation of households. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | three baseline optimization problems for comparison with our value-stacking model. We also introduce the performance metrics for investigating the impact of prediction errors by measuring the extra cost based on the rolling-horizon optimization. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | To evaluate the contribution of each value stream, we study a set of baseline optimization problems. Specifically, we con- sider V2G, V2H, and EV energy trading alone, respectively, as three baseline problems. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | EVd = pu,t V2H. • For EV energy trading alone, prosumers with EVs trade energy with others by scheduling variables, including grid, pu,t pu,t EVd = pu,t Sell, and pu,t Sell. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | Moreover, to evaluate the marginal contribution of each value stream in our value-stacking problem presented in Sec- tion III-C, we consider three additional baseline problems. Each baseline problem excludes one value stream from the value-stacking problem with all value streams in Section III-C, showing the marginal contribution of each excluded value stream. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | where Cu,t cost represent the cost of prosumer u in time slot t based on real-world data and prediction data, respectively. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | We evaluate our proposed network-aware value-stacking on the IEEE 33-bus test system, depicted in Fig. 5, where three communities connect to nodes 4, 25, and 32, respectively. The total load of the IEEE 33-bus is 3.72MW, and the base case values for the apparent power and voltage are 10MVA and 12.66kV, respectively. In each community, there are 20 prosumers (i.e. 20 households, each of which has a PV system and an EV). The EV battery capacity for each prosumer is 50kWh, the initial stored energy ranges from 20kWh to 30kWh, and the maximum charging rate is 7kW. We use real market data from NEM, ISO-NE, NY-ISO, and evaluate two tariffs for prosumers, e.g., TOU and TPT. For the TOU tariff, the off-peak price is AU$ 0.2, and the peak price is AU$ 0.32. For the TPT tariff, the energy price is AU$ 0.2, and the peak price is AU$ 0.8 [22]. For the TOU plan in NY-ISO and ISO-NE, the off-peak price is US$ 0.02, the shoulder price is US$ 0.11, and the peak price is US$ 0.321. The value-stacking model is validated in MATLAB using the YALMIP toolbox [23] on a PC with Intel(R) Core(TM) i7-7700HQ CPU and 32GB RAM. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | week, including five working days and two holidays. The results are presented in Fig. 6. The vertical axis represents the total cost reductions in % for value stacking and three baselines, including V2G, V2H, and EV energy trading alone, respectively. The two clusters of bars on the horizontal axis correspond to two tariffs, in which TOU and TPT are applied to prosumers, respectively. Under TOU pricing, value stacking achieves a cost reduction of 12.50%. Among three value streams, V2G creates the lowest cost reduction of 1.40%. The benefit of V2H is higher than energy trading in the local market since all prosumers have the same peak load hours, which leads to fewer opportunities for energy trading but more contribution of V2H in the cost reduction. Under TPT, value stacking achieves a cost reduction of 11.75%, which is lower compared to that under TOU pricing. Energy trading contributes more cost reduction than V2H because prosumers’ peak loads can appear at diverse times and energy trading among prosumers can enable better peak shaving. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | peak price of the TOU pricing is much lower than the peak and shoulder prices of the tariff, prosumers prefer to store energy in off-peak hours and sell energy to the local market (for energy trading) when the TOU tariff price is high in peak hours. Hence, energy trading benefits prosumers more in both NY-ISO and NE-ISO. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | the household load and PV generation for prosumers, we utilize the LSTM network model in III-B to predict hourly load and PV generation for 60 prosumers, respectively, over 200 days. Specifically, we adopt 12 months of data (from Jan. 1, 2020, to Dec. 31, 2020) of load and PV generation for prosumers as the training datasets, where the loss function is the root mean square error (RMSE). The number of hidden units is 400, the initial learning rate is 0.005, the maximum epochs is 250, and the Adaptive Moment Estimation (Adam) algorithm is utilized to train the LSTM model. Then, we forecast 200-day data (from 1/1/2021 to 19/7/2021) using the trained models. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | Fig. 10. Relationship between the mean extra cost rate and the relative error of load prediction. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | To evaluate the performance of the LSTM model, we randomly choose four prosumers during one week (i.e., 168 hours) to compare the load forecast with real load in Fig. 8. According to Eq. (29), the relative errors in the predicted load for four prosumers are 0.47, 0.43, 0.74, and 0.36, respectively. For four prosumers, the prediction is not good enough in peak load hours, as it is challenging to capture the behavior of individual households. Prosumer 3 in Fig. 8 (c) has the largest relative error among the four prosumers since the error of forecast in peak load hour is large. Similar to load forecast, we also show the PV generation prediction for these four prosumers depicted in Fig. 9. Based on Eq. (30), we compute the relative errors of the predicted PV generation for four prosumers and the results are 0.25, 0.21, 0.26, and 0.22, respectively. Compared with load forecast, PV generation prediction achieves better performance, using the same LSTM architecture, which indicates that capturing the PV generation is relatively easier than household energy consumption behav- ior. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | Fig. 11. Relationship between the mean extra cost rate and the relative error of PV generation prediction. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | prediction errors is smaller. Specifically, when the relative PV prediction error achieves 30 ∼ 35%, the mean extra cost rate is 4.4% ranging [2.4, 7.2%]. The results are obtained from our simulation setting and can be explained as follows. Sometimes, EVs are not at home in the daytime when the PV systems generate energy, and thus the impact of PV generation prediction errors on EV value-stacking is smaller. |
Processed_Network-Aware_Electric_Vehicle_Coordination_for_Ve.txt | In this paper, we proposed a V2X value-stacking optimiza- tion problem for EV coordination considering local network constraints to maximize the economic benefits. We formulated a RHO problem to solve the value-stacking problem in real- time using predicted PV generation and load information. We further evaluated the impact of prediction errors in load and PV generation on our proposed EV value-stacking problem. The results showed that the value-stacking model achieved better performance with a cost reduction of 12.50%, when TOU was applied as the retail tariff. The marginal contributions of three value streams in three markets, namely NEM, ISO- NE, and NY-ISO, were also studied. The results demonstrated that energy trading contributed the most to cost reduction in three markets. The numerical results revealed that the mean extra cost was sensitive to the relative error in load prediction, while PV generation prediction error has a smaller impact. Our observation indicates that more accurate load prediction is needed for future EV value stacking models or systematic approaches are required to mitigate the impact of load prediction errors in value stacking optimization. |
Processed_On_the_consistency_of_neutron-star_radius_measurem.txt | Independent of the issues for measurement, are un- certainties about precisely how the neutron star’s at- mosphere affects the emerging radiation. Although the burst spectra are typically found observationally to be consistent with a blackbody (e.g. Swank et al. 1977; Kuulkers et al. 2002), scattering effects have long been understood to distort the spectrum sufficiently to bias the measured temperature (e.g. London et al. 1984, 1986). This distortion is usually parameterised via a spectral distortion factor fc = Tbb/Teff where Tbb is the measured blackbody (or colour) temperature, and Teff is the effective temperature of the atmosphere. Most recent work adopt a narrow range of fc = 1.3–1.4 at burst lumi- nosities well below the Eddington limit (e.g. Madej et al. 2004), and often neglect the variations in fc that may arise during the burst (e.g. Suleimanov et al. 2011b). A more rigorous approach involves fitting the observed vari- ation in the blackbody normalisation (in response to the changing fc) as a function of flux (e.g. Suleimanov et al. 2011a), although the predicted model curves cannot yet reproduce the range of observed behaviour. |
Processed_On_the_consistency_of_neutron-star_radius_measurem.txt | using a uniform, homogeneous sample of bursts, from the low-mass X-ray binaries GS 1826−24 and KS 1731−26. We use Rossi X-ray Timing Explorer (RXTE) Propor- tional Counter Array (PCA) data to test for intrinsic systematic effects which might influence burst normali- sation measurements beyond any additional effects which might arise from variations in the ignition conditions and fuel composition. In a related paper, Zamfir et al. 2011 (submitted to ApJ; hereafter Z11) used the same data to infer the mass and radius of GS 1826−24. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.