Ungulates selectively browse on young native seedlings preventing indigenous ecosystems from regenerating naturally this browsing can lead to an altered state An ungulate control operation will take place in the Maitai and Roding catchments from 12-23 May essential for preserving and restoring biodiversity in these areas The Maitai and Roding catchments are vital ecosystems in Nelson the presence of ungulates—browsing mammals like deer and pigs—poses a significant threat to these areas “The upcoming ungulate cull represents a critical step in our ongoing commitment to protect and enhance Nelson’s unique biodiversity,” says Group Manager Environmental Management Mandy Bishop “Populations of deer and other hoofed mammals are having a devastating impact on our native forest understory preventing regeneration and threatening the delicate balance of our local ecosystems By responsibly managing these introduced species we’re creating space for our indigenous flora to thrive which in turn supports our native bird and insect populations Regular culling operations align with our broader environmental strategy and demonstrates our commitment to being effective kaitiaki of our natural heritage.” By controlling ungulate population numbers we can limit their impact on important indigenous ecosystems reducing the impact of the ungulate population protects Council and community assets A key aspect of the ungulate control operation is the partnership with Ngāti Koata and Tasman Pine This collaboration allows for coordinated control efforts across neighbouring areas reducing the risk of reinvasion and ensuring a more comprehensive approach to pest management Ngāti Koata’s involvement also includes paying for the retrieval of meat from culled animals which is then distributed within the iwi community The next cull is scheduled to take place from 12 – 23 May (with the weekend of 17-18 May open for recreation users) and will be carried out by contractors Trap and Trigger Trap and Trigger use both aerial (helicopter) and ground control methods The contractor will implement measures to minimise the impact of noise on residents though some minor disruptions may still occur there will be area and trail closures to ensure safety These closures will be clearly marked at trailheads Front country hunting including Coppermine Trail will take place in: including Coppermine Trail will take place in: These reserves and areas will be closed during weekdays but will reopen to all users during weekends from midnight Friday till 11:59pm Sunday For further information and FAQs please go to the Shape Nelson page: https://shape.nelson.govt.nz/ungulate-management Researchers and social practitioners working with people from refugee and migrant backgrounds in Aotearoa say Tino Rangatiratanga (Self Determination) should […] A researcher investigating the experiences and perceptions of New Zealand born Pacific youth says it’s important for adults to support […] Subscribe to free Weekly Community Scoop updates Subscribe for Daily Community Scoop news updates delivered to your inbox: London: A previously neglected scrap of land has been transformed for nature I visit to immerse myself in the botanical throng The directions I’m following were detailed, but unconventional for a London address: “A river; a winding path; a gate; an apple tree; a reedbed; a plank … Stay on the left as it’s wobbly.” What my friend Paul Powlesland didn’t say is that the plank would feel like a modern day Sweet Track a walkway raised above the tidal marsh into another world Arriving after midnight at his mooring on London’s neglected third river I find myself in a susurrating darkness that almost subsumes the late-night traffic noise that the final section of plank would be slimy with rain and listing precariously above the water The path to the yurt Photograph: Amy Jane BeerLondon is out there The sense of being surrounded calls to my drowsy mind a line in Maurice Sendak’s Where the Wild Things Are: “That night … the walls became the world all around.” I wake to bright sunshine and a blast of Cetti’s warbler song – as loud and assertive as the wren I hear at home The yurt door opens on to a wall of whispering green Over the reedheads I can see the breeze-ruffled tops of aspen I start to explore the thatch of vegetation along the riverside path – an eclectic mix including nettle whose nicknames include river wormwood (apt) and naughty man (intriguing) are the deadly umbels and fronds of hemlock water dropwort The diversity is unexpected, but not accidental. Paul is a co-founder of the Lawyers for Nature collective and the founder of the River Roding Trust “The Trust took on this scrap of land for nature restoration The plan is to have one of every British native tree here so anyone can come and meet them.” The reeds applaud Country diary is on Twitter at @gdncountrydiary The options for members of a UK jury are fairly standard: ‘by Almighty God’ Paul Powlesland chose none of these when he was called to jury service in July the 38-year-old barrister opted for the River Roding “from her source in Molehill Green” to her “confluence with the Thames” Powlesland admits that the judge was a “little perplexed” at first This was partly because Powlesland also declared the juror’s secular ‘affirmation’ (“to make sure the legal boxes were ticked”) but mostly because the judge believed the sincerity of the river lover’s position “I explained that I believe the river to be sacred but also I could demonstrate that this wasn’t just something I was saying but that I do care for the river constantly in my spare time,” says Powlesland When not donning his barrister’s gown and advocating in court (usually on housing or employment issues) he’s dredging rubbish from the River Roding on his canoe or planting trees or placing benches along its banks As founder of the River Roding Trust Powlesland has made it his business to protect and help regenerate the 31-mile river which rises in a patch of rural Essex and then flows west towards London travelling under the M25 before ending up in the heavily urbanised borough of Barking and Dagenham The latter provides the mooring point for the 45-foot narrow boat that Powlesland has called home for the last seven years Pretty as the reed beds that immediately surround his barge are the barrister admits that the kilometre-or-so long stretch where he concentrates most of his conservation efforts is “on a knife edge” Something magic happens when you get to know a specific part of nature – and there’s a motorway fewer than 100 metres away the pollution from excessive pesticide and fertiliser use upriver he still believes the river is “redeemable” From a man who earns his living from his command of language He doesn’t come from a family of environmentalists nor have rivers always grabbed his heart: his is a gradual passion born from years of living on the water (before Barking he spent five years on the canals of east London) You can decide right now to become a guardian of your local river and to look after and care for it on that basis His eventual answer comes more as assertion than explanation: “Something magic happens when you get to know a specific part of nature and that’s the Roding.” Just as every river is different so are the effects they have on different people But what all rivers have in common is their longevity: almost all predate us an inversion of the human-centric narrative that we tell ourselves Find a river nearby and just spend time on this time spent just instils a connection and love.” And from that love springs an equally natural desire to care for it Through Lawyers for Nature Powlesland is pushing for the UK to formally grant rivers a legal right of protection The jury is still out on whether politicians will act he says: more important is people power: “You can decide right now to become a guardian of your local river and to look after and care for it on that basis.” Support Positive News Website: Atomic Smash From our cover story about joyful jobs, to reader reflections on abundance – the uplifting new issue of Positive News magazine is a thought-provoking read to brighten your spring. Volume 8 - 2021 | https://doi.org/10.3389/fmats.2021.786502 This article is part of the Research TopicVirtual Materials DesignView all 14 articles Effective properties of functional materials crucially depend on their 3D microstructure we investigate quantitative relationships between descriptors of two-phase microstructures consisting of solid and pores and their mass transport properties we generate a vast database comprising 90,000 microstructures drawn from nine different stochastic models and compute their effective diffusivity and permeability as well as various microstructural descriptors this is the largest and most diverse dataset created for studying the influence of 3D microstructure on mass transport we establish microstructure-property relationships using analytical prediction formulas artificial (fully-connected) neural networks this is the first time that these three statistical learning approaches are quantitatively compared on the same dataset The diversity of the dataset increases the generality of the determined relationships and its size is vital for robust training of convolutional neural networks their structural descriptors and effective properties as well as the code used to study the relationships between them available open access virtual materials testing is an approach of increasing importance where mathematical models are used for both the analysis of artificially generated materials structures as well as for the investigation of their effective properties By systematically varying the model parameters a large number of virtual but realistic 3D microstructures can be drawn from stochastic models just at the cost of computer simulations These structures serve as geometry input for spatially-resolved numerical simulations of effective properties together with the computation of various descriptors of the 3D microstructures quantitative microstructure-property relationships can be established although in many cases with small datasets and/or only with 2D structures CNNs have the tendency to be even less transparent than ANNs in terms of understanding how the prediction works The rest of this paper is organized as follows In Section 2 the definitions of various structural descriptors are explained as well as their estimation from 3D image data They are used as input variables for establishing microstructure-property relationships namely diffusivity and permeability of the pore space Section 4 introduces the stochastic models used for the artificial generation of 3D microstructures whereas in Section 5 three different approaches are explained which are applied to establish the microstructure-property relationships a discussion of the results obtained in this paper is provided in Section 6 Note that all microstructures that are drawn from these stochastic models fulfill periodic boundary conditions in x- This is taken into account when estimating the structural descriptors as presented below In addition to simple scalar descriptors of Ξ such as volume fraction we also consider more complex descriptors like the chord length distribution the spherical contact distribution and the distribution of geodesic tortuosity we represent these distributions by their quantiles starting from 5%- up to 95%-quantiles in 5% steps To begin with, we consider one of the simplest but most important structural descriptors, namely the porosity ɛ ∈ [0, 1], i.e., the volume fraction of the random pore space Ξ⊂R3, where ε=E(ν3(Ξ∩[0,1]3)) and ν3 denotes the three-dimensional Lebesgue measure. This characteristic can be easily estimated from 3D image data by the point-count method (Chiu et al., 2013) Note that this estimation method has obviously not to be adapted further to account for periodic boundary conditions where the transport direction is from low x-values to high x-values Note that the transport paths are allowed to “leave” the sampling window in the y- and z-directions in order to account for periodic boundary conditions where it is straightforward to apply periodic boundary conditions in y- and z-directions Note that due to our stationarity and isotropy assumptions the values of MIP do not depend on the choice of the predefined direction where the intrusion starts at low x-values the radius rmin is defined as the radius for which MIP(rmin) equals half the porosity The chord length distribution of the pore space, which is modelled by a stationary random set Ξ⊂R3, is defined as follows (Ohser and Mücklich, 2000; Ohser and Schladitz, 2009; Chiu et al., 2013) Given a predefined direction φ∈[0,π2]×[0,2π) the chord length distribution of the random set Ξ in direction φ is the distribution of the length L of the so-called typical line segment (selected at random) in Ξ ∩ ℓ where ℓ denotes the line passing through the origin in direction φ The distribution of L is denoted by d(L) and Note that d(L) does not depend on the particular choice of φ when considering stationary and isotropic random sets The mean and the standard deviation of the chord length distribution are denoted by m(L) and σ(L) observing Ξ ∩ ℓ ∩ W within some sampling window W⊂R3 the chord length distribution is estimated by counting subsequent voxels belonging to the pore space along the x-axis and computing the empirical distribution function of the lengths of these voxel sequences Note that periodic boundary conditions can be simply accounted for by merging the first and the last chord in Ξ ∩ ℓ ∩ W provided that both chords belong to the pore space Consider the (random) distance H from the typical point of Ξc=R3\Ξ to the nearest point within Ξ 1] with FH(r)=P(H≤r) for each r ≥ 0 is called the spherical contact distribution function of Ξ where B(o,r)⊂R3 denotes the closed ball with radius r centered at the origin (Chiu et al., 2013). The mean, standard deviation and distribution of H are denoted by m(H), σ(H) and d(H), respectively. These quantities can be estimated from voxelized 3D image data using the algorithm proposed by Mayer (2004) which relies on the computation of the Euclidean distance transform periodic boundary conditions are taken into account by computing the Euclidean distance transform with respect to periodic boundary conditions as described above in Section 2.4 As in case of the distribution of geodesic tortuosity and the chord length distribution the distribution d(H) is represented by 19 quantiles For a stationary and isotropic random set Ξ⊂R3 describing the pore space of a porous material, the two-point (pore-pore) correlation function C: [0, ∞) → [0, 1], which is also called covariance function (Serra, 1982; Ohser and Schladitz, 2009) where x∈R3 is an arbitrary point with distance r to the origin (Matheron, 1975; Torquato, 2002; Chiu et al., 2013). This quantity can be estimated from voxelized image data by the Fourier method described in Ohser and Schladitz (Ohser and Schladitz, 2009) where no further step is required to account for periodic boundary conditions we represent the two-point correlation function by C(0) where 167=⌈3⋅(0.5⋅192)2⌉ is the maximal distance of two points within the sampling window W = [0,192]3 with respect to periodic boundaries Note that the pore-solid and solid-solid correlation functions are uniquely determined by the pore-pore correlation function In this section we briefly explain two effective transport properties for which numerical simulations are carried out to estimate these quantities from 3D image data Effective tortuosity of the pore space is usually defined by effective diffusivity is obtained by numerically solving Laplace’s equation on Ξ the following second-order differential equation is solved: where c denotes the concentration of the diffusing species Apart from mass conservation within the pore space one has to ensure that the diffusing species can not intrude into the solid phase which is formally described the following equation at the interface: where the outward pointing unit normal is denoted by n and ◦ denotes the scalar product the following equations are the driving force for the flux in x-direction: where x0 and xmax denote the two parallel planes described by x = 0 and x = 192, respectively. Note that periodic boundary conditions are applied in y- and z-direction. Further technical details regarding the implementation of the equations above can be found in (Cooper et al., 2016) The M-factor, defined as M = Deff/D0, is now given by M = ɛ/τeff, where it holds that M ∈ [0, ɛ] and, equivalently, τeff ≥ 1, according to Eq. 21.14 in the book of Torquato (2002) lower values of M correspond to more pronounced transport limitations whereas a high value of M indicates nearly no hindrance of diffusion processes Δp is the applied pressure difference and d is the length of the microstructure in the flow direction The permeability is independent of the fluid and the pressure difference and is hence a property solely of the microstructure provided that the Reynolds number is sufficiently small (<0.01) This also ensures that the velocity is proportional to the pressure difference since we are dealing with simulated microstructure data on the voxel grid computed permeabilities are given in (voxel unit)2 the microstructure models considered in the present paper are designed in such a way that the resulting sets of artificially generated microstructures are disperse in the sense that their microstructure descriptors cover a wide spectrum of values keeping the value of a certain microstructure descriptor fixed the values of other characteristics can still be varied “independently” (to a certain extent) note that due to inherent correlations between some pairs of geometric microstructure descriptors the space of values that can be covered is naturally limited porosity values close to one typically go along with a low mean geodesic tortuosity the 90,000 microstructures drawn from the nine different stochastic models lead to an extensive dataset representing a broad range of morphologies which allows us to attribute a certain generality of the microstructure-property relationships determined in the present paper Examples of the different types of microstructures showing an artificially generated fiber system (I) as well as systems of hard ellipsoids (VI) Note that the solid phase is always depicted in blue whereas mass transport takes place in the transparent porous phase Specific surface area (in (voxelunit)−1) (A) standard deviation of geodesic tortuosity (C) characteristic bottleneck radius rmin (in voxelunit) (D) constrictivity (E) and mean chord length (in voxelunit) (F) as a function of porosity where 250 structures have been randomly selected for each of the nine different microstructure models Note that the roman numbers in the legend correspond to the numbering of the nine stochastic model types mentioned at the beginning of Section 4 Systems of channels are generated using essentially the same code as for the fiber systems considered in Section 4.1 treating the fibers as porous channels instead and the fiber radii are sampled in the same fashion as in Section 4.1 Starting from a fully solid structure (without any channels) new channels are added until the desired porosity is reached where spheres instead of random ellipsoids are considered Configurations of smoothed soft ellipsoids are generated in the same manner as the systems of soft ellipsoids considered in Section 4.8, with the only difference that the final discretized structure is smoothed with a Gaussian filter, the standard deviation of which is randomly sampled in the range of [2, 16] voxels (Russ, 2007; Gonzalez and Woods, 2008) For the 90,000 samples of 3D microstructures generated by the stochastic models described in Section 4 the structural descriptors explained in Section 2 as well as the effective transport properties stated in Section 3 are computed This data is then used to establish microstructure-property relationships by means of three different approaches (artificial) fully-connected neural networks (ANNs) and convolutional neural networks (CNNs) the data is randomly shuffled and split into three subsets of training This is done in a stratified manner such that an equal number of microstructures of each type is included in each of the three datasets which is the same for each of the three types of prediction models is 70% training data (7,000 per type of microstructure and 63,000 in total) and 15% each for the validation and test data (1,500 per type of microstructure and 13,500 in total) Several error measures are considered to assess predictive performance yk is the ground truth data of a given output variable and ŷ1,…,ŷk are the corresponding estimates predicted by the model the MSE loss is used either on the output variables themselves or on transformed outputs Further details can be found below in Sections 5.1–5.3 for the three types of microstructure-property relationships considered in this paper Note that the MSE loss is an appropriate loss function for optimization because it is differentiable the mean absolute percentage error (MAPE) loss is used because the MAPE loss is not everywhere differentiable it is less well-behaved as an optimization target but on the other hand it is more interpretable we consider the coefficient of determination R2 ∈ [0 where ȳ denotes the empirical mean of the ground truth data y1 Since the values for permeability cover several orders of magnitude the value of R2 would be dominated by those terms that involve large values we also consider the coefficient of determination on the log scale where ȳlog is defined as the empirical mean of the sample log(y1) Note that the values for the effective tortuosity are always larger than for the M-factor which is probably caused by the fact that τeff already contains the porosity ɛ it is worth mentioning that a low value of the functional dependence measure does not automatically imply that this quantity should not be considered for predicting a certain effective property y) only contains some information on the predictive power of x with regard to y but not regarding the usefulness of x in combination with other structural descriptors This also explains the fact that the normalized quantity M turned out to be predicted with higher accuracy than τeff and therefore we stick to M for the rest of this work It is also interesting to point out that m(L) and σ(L) seem to be closely related to permeability and effective tortuosity considering that the chord length distribution is rarely used in the literature for establishing microstructure-property relationships y) between a scalar structural descriptor x and an effective property y Since there are no hyperparameters in case of analytical prediction formulas we merge the training set with the validation set for computing the fitting parameters and use the test set for assessing performance we state nine analytical prediction formulas and their fitting parameters a comparison of these formulas is carried out including an interpretation of the results where the additional constraint c1 + c2 ≥ 0 is used to ensure that M̂∈[0,1]. This leads to c1 = 1.25, c2 = − 1.25 and c3 = − 7.82. Finally, we consider a formula for predicting the M-factor by porosity as well as the mean and standard deviation of geodesic tortuosity of the pore space, see Barman et al. (2019): where least-squares fitting gives that c1 = 1.18 Having discussed parametric formulas for predicting the M-factor we now predict the permeability κ using geometric microstructure descriptors given in Section 2 Since the values of κ can cover several orders of magnitude the fitting of parameters is carried out on the log scale which has been introduced in Neumann. et al. (2020). By least-squares regression, we obtain that c1 = 0.16, c2 = 2.05, c3 = 0.64 and c4 = − 7.31. Moreover, we consider still another type of a parametric prediction formula for κ, proposed in Neumann. et al. (2020) where least-squares fitting gives that c1 = 0.24, c2 = 0.92, c3 = 0.08, c4 = 1.6 and c5 = − 6.82. Note that for fitting the parameters in Eq. 14 we use the additional constraint c2 we use a convex combination of rmin and rmax which is subsequently squared to ensure the right unit of permeability a further parametric prediction formula for κ which has been discussed in the literature see Röding et al. (2020) where least-squares regression leads to c1 = 0.14 c2 = 2.07 and c3 = − 8.57 In addition to the results mentioned above we consider the following prediction formula for κ: which uses the constrictivity β within the exponent of the porosity ɛ, similar to Eq. 11 where the fitting parameters are given by c1 = 0.14 c3 = − 1.38 and c4 = − 7.37 we predict κ by the porosity ɛ the mean geodesic tortuosity m(τgeo) and the median rmin via where least-squares fitting on the log scale gives that c1 = 0.25 c2 = 1.6 and c3 = − 6.6 we consider a prediction formula for κ which involves the mean chord length m(L) of the pore space: where we use the constraint c2 + c5 = 2 in order to obtain the right unit (voxels2) and the fitting parameters are given by c1 = 0.1 c5 = 1.37 and c6 = − 6.74 FIGURE 3. Prediction of effective transport properties using analytical formulas. Top row (from left to right): Prediction of M-factor via Eqs. 10–12. Middle row (from left to right): Prediction of permeability via Eqs. 13–15. Bottom row (from left to right): Prediction of permeability via Eqs. 16–18 Note that the scatter plots show results based on the test data TABLE 2. Error measures computed on the test set, corresponding to the analytical prediction formulas Eqs. 10–18 for M-factor and permeability Note that in case of predicting permeability the quantity Rlog2 denotes the coefficient of determination on the log scale the outputs are transformed in the following fashion the logit-transformed M-factor y = log(M/(1 − M)) and the log-transformed permeability y = log(κ) are used as the target outputs This yields the benefit that the inverse-transformed predictions belong to (0 Descriptor sets used as input for the ANNs together with the dimensions of the corresponding input vectors 9 and 10 involve four scalar quanities and one distributional characteristic described by 19 quantiles whereas the two-point correlation function in Model 1 is evaluated for 168 different radii it turned out that the learning rate (LR) has considerable impact on the results we design an LR scheme with a step-wise increasing and then step-wise decreasing learning rate 10–2.5} for 1,000 epochs (iterations over the whole training set) each the training procedure comprises 25,000 epochs 100 networks are trained using different random seeds Note that the random seed controls the weight initializations in the network as well as the shuffling of data in the SGD The model yielding the minimal validation loss (over all epochs and all runs) for each set of microstructural descriptors is selected the average execution time for each run is 3.7 h FIGURE 4. Illustration of the ANN architecture with 4 hidden layers, each with 64 nodes, where only a smaller number of nodes is shown in this figure for clarity. Furthermore, the input in this figure is 5-dimensional, but in the present paper the input dimension varies from 3 to 236, see Table 3 Scatter plots visualizing the prediction results for M-factor and permeability are shown in Figures 5, 6, respectively. Furthermore, error measures for the prediction of M-factor and permeability are shown in Tables 4 and 5 It turns out that the predictions obtained by ANNs are consistently better than those obtained by the analytical prediction formulas considered in Section 5.1 when the same descriptors are used as input adding more complex descriptors like quantiles (of tortuosity the best results are obtained using all the computed descriptors Top row (from left to right): Prediction of M via Models 1–3 Second row (from left to right): Prediction of M via Models 4–6 Third row (from left to right): Prediction of M via Models 7–9 Bottom row (from left to right): Prediction of M via Models 10–12 Top row (from left to right): Prediction of κ via Models 1–3 Second row (from left to right): Prediction of κ via Models 4–6 Third row (from left to right): Prediction of κ via Models 7–9 Bottom row (from left to right): Prediction of κ via Models 10–12 Error measures for the prediction of M via ANNs where MSE is given for the training Note that MSE is evaluated on the logit scale and MAPE on the linear scale Error measures for the prediction of κ via ANNs the ANN model involving the two-point correlation function performs better than the model (no 8) involving the distribution of tortuosity Note that the Models 3 and 4 which do not use neither rmin nor the specific surface area S perform substantially worse than all other models with regard to permeability including all descriptors (Model 12) yields the best performance for predicting both M and κ comparing models with low-dimensional (Models 1–7) and high-dimensional (Models 8–12) descriptor sets the best low-dimensional descriptor (Model 7) gives very good performance going from 5 to 236 input dimensions reduces the MAPE from 2.74 to 2.03% for M and from 7.47 to 6.51% with regard to κ Considering that this reduction in error requires a massive reduction in interpretability of the model it is not obvious how to make this trade-off between model complexity and performance the main building blocks are convolutional layers A typical CNN architecture comprises convolutional layers the input is convolved with several convolution kernels The convolutions themselves are linear operations but a nonlinear activation function f:R→R is applied to the result to produce the outputs the feature maps are downsampled by computing the mean or maximum on small (typically non-overlapping) patches of the feature maps from the preceding layer reducing the resolution such that the next convolutional layer can extract information from another spatial scale After the convolutional and pooling layers fully-connected layers are typically used to obtain a scalar output The first part of the CNN can be thought of as a feature extractor that produces geometrical features which are qualitatively similar to the ones used for the analytical prediction formulas and ANNs considered in the previous sections of this paper whereas the second part corresponds directly to the ANNs themselves are fed into the convolutional part of the network each in turn consisting of two convolutional layers (the numbers of filters are indicated in the figure) followed by an average pooling layer The feature maps produced as output from the convolutional part are passed to 4 fully-connected layers with 64 nodes each the average execution time is 140 h The model yielding the minimal validation loss over all epochs is selected Because of the large computational workload for CNNs we do only one run for M-factor and one for permeability In addition to this ordinary CNN, we train the same architecture with a different kind of input data. Instead of using the structure arrays, we compute the Euclidean distance transform in the pore space which effectively comprises a spatial map of local pore sizes (Russ, 2007) this approach of using the distance transform as a representation of the pore space has not been used before as inputs to a CNN where these inputs are again rescaled to 963 arrays The only differences compared to the ordinary CNN are that the input data has to be stored in 32-bit floating point precision (leading to high demands in storage space and test) and that the data are batch-wise rescaled by a factor 1/24 (the 95%-quantile of the distance transform values is approximately 24) Scatter plots visualizing the prediction results for M-factor and permeability are shown in Figure 8. Furthermore, error measures are shown in Table 6 for both M and κ and for both the ordinary CNN (briefly denoted by CNN) and the distance-transform CNN (denoted by DT-CNN) all CNNs perform better than their best ANN counterparts the best models attaining 1.65% (M-factor) and 3.78% (permeability) MAPE although this is at the expense of even less interpretability than for the highest-dimensional descriptor used for the ANNs the differences between the ordinary CNN and the DT-CNN are not substantial and neither one of them is consistently better This is possibly because the ordinary CNN learns similar information as that already supplied to the DT-CNN we conclude that given the increased computational workload of the distance transform there are at least no substantial benefits of using the DT-CNN over the ordinary CNN showing (A) prediction of M using an ordinary CNN (C) prediction of κ using an ordinary CNN and (D) prediction of κ using a DT-CNN Error measures for the prediction of M and κ via ordinary CNN and DT-CNN where MSE is given for the training Note that MSE is evaluated on the logit scale for M and log scale for κ The microstructures, descriptors, and the code used to study microstructure-property relationships are available open access via the following Zenodo repository: https://zenodo.org/record/4047774, see Prifling et al. (2021c) MR and MN generated the data and performed the analysis of microstructure-property relationships PT developed the microstructure model for fiber and channel systems All authors discussed the results and contributed to writing the paper The financial support of the Swedish Research Council for Sustainable Development (grant number 2019-01295) and the Swedish Research Council (grant number 2016–03809) is acknowledged the presented work was financially supported by the Bundesministerium für Bildung und Forschung (BMBF) within the project HiStructures under the grant number 03XP0243D as well as by the Deutsche Forschungsgemeinschaft (DFG) under the grant number SCHM997/39-1 The computations were in part performed on resources at Chalmers Centre for Computational Science and Engineering (C3SE) provided by the Swedish National Infrastructure for Computing (SNIC) A GPU used for part of this research was donated by the NVIDIA Corporation The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher Victor Wåhlstrand Skärström is acknowledged for assistance with implementing the convolutional neural networks Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., et al. (2015). TensorFlow: Large-Scale Machine Learning on Heterogeneous Systems. Available at: tensorlow.org (Accessed September 29 Google Scholar Morphological Modelling of Three-phase Microstructures of Anode Layers Using SEM Images CrossRef Full Text | Google Scholar Quantifying the Influence of Charge Rate and Cathode-Particle Architectures on Degradation of Li-Ion Cells through 3D Continuum-Level Damage Models CrossRef Full Text | Google Scholar Deep Learning-Driven Permeability Estimation from 2D Images CrossRef Full Text | Google Scholar Prediction of Diffusive Transport through Polymer Films from Characteristics of the Pore Geometry CrossRef Full Text | Google Scholar Modeling Phenomena of Flow and Transport in Porous Media Theory and Applications of Transport in Porous Media Google Scholar Random Search for Hyper-Parameter Optimization Google Scholar Bounds on Fluid Permeability for Viscous Flow through Porous media CrossRef Full Text | Google Scholar Modified Collective Rearrangement Sphere-Assembly Algorithm for Random Packings of Nonspherical Particles: Towards Engineering Applications CrossRef Full Text | Google Scholar Julia: A Fresh Approach to Numerical Computing CrossRef Full Text | Google Scholar Statistical Investigation of Structural and Transport Properties of Densely-Packed Assemblies of Overlapping Spheres Using the Resistor Network Method CrossRef Full Text | Google Scholar “Large-scale Machine Learning with Stochastic Gradient Descent,” in Proceedings of COMPSTAT’2010 CrossRef Full Text | Google Scholar Chapman & Hall/CRC Handbooks of Modern Statistical Methods Google Scholar A Trust Region Method Based on Interior Point Techniques for Nonlinear Programming CrossRef Full Text | Google Scholar Google Scholar Google Scholar CrossRef Full Text | Google Scholar “Fast and Accurate Deep Network Learning by Exponential Linear Units (Elus),” in 4th International Conference on Learning Representations Google Scholar An Interior Trust Region Approach for Nonlinear Minimization Subject to Bounds CrossRef Full Text | Google Scholar On the Convergence of Interior-Reflective Newton Methods for Nonlinear Minimization Subject to Bounds CrossRef Full Text | Google Scholar TauFactor: An Open-Source Application for Calculating Tortuosity Factors from Tomographic Data CrossRef Full Text | Google Scholar Microstructural Analysis of an LSCF Cathode Using In Situ Tomography and Simulation CrossRef Full Text | Google Scholar Approximation by Superpositions of a Sigmoidal Function CrossRef Full Text | Google Scholar An Introduction to the Theory of Point Processes I Google Scholar An Introduction to the Theory of Point Processes II Google Scholar PARROT - A Pilot Study on the Open Access Provision of Particle Discrete Tomographic Datasets Google Scholar “Benchmarking Materials Property Prediction Methods: The Matbench Test Set and Automatminer Reference Algorithm CrossRef Full Text | Google Scholar Stochastic 3D Modeling of the Microstructure of Lithium-Ion Battery Anodes via Gaussian Random fields on the Sphere CrossRef Full Text | Google Scholar Artificial Generation of Representative Single Li-Ion Electrode Particle Architectures from Microscopy Data CrossRef Full Text | Google Scholar Quantitative Relationships between Microstructure and Effective Transport Properties Based on Virtual Materials Testing CrossRef Full Text | Google Scholar Curves and Surfaces in Geometric Modelling: Theory and Algorithms Google Scholar A Lattice Boltzmann Method for the Advection-Diffusion Equation with Neumann Boundary Conditions CrossRef Full Text | Google Scholar Investigation of the Effect of the Tortuous Pore Structure on Water Diffusion through a Polymer Film Using Lattice Boltzmann Simulations CrossRef Full Text | Google Scholar Tortuosity in Porous Media: A Critical Review CrossRef Full Text | Google Scholar Study of Simple Hydrodynamic Solutions with the Two-Relaxation-Times Lattice Boltzmann Scheme Google Scholar “Understanding the Difficulty of Training Deep Feedforward Neural Networks,” in Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics Google Scholar Google Scholar and Tortuosity of Porous media from Images by Deep Learning PubMed Abstract | CrossRef Full Text | Google Scholar “Influence of Conductive Additives and Binder on the Impedance of Lithium-Ion Battery Electrodes: Effect of Morphology CrossRef Full Text | Google Scholar Stochastic Microstructure Modeling and Electrochemical Simulation of Lithium-Ion Cell Anodes in 3D CrossRef Full Text | Google Scholar Hernández-García Further Advantages of Data Augmentation on Convolutional Neural Networks Maglogiannis (Cham: Springer International Publishing) CrossRef Full Text | Google Scholar Effective Diffusion Coefficients in Random Packings of Polydisperse Hard Spheres from Two-point and Three-point Correlation Functions CrossRef Full Text | Google Scholar Review of Tortuosity: Classical Theories and Modern Methods Google Scholar The Next Level in Chemical Space Navigation: Going Far beyond Enumerable Compound Libraries PubMed Abstract | CrossRef Full Text | Google Scholar The Influence of Constrictivity on the Effective Transport Properties of Porous Layers in Electrolysis and Fuel Cells CrossRef Full Text | Google Scholar Multilayer Feedforward Networks Are Universal Approximators CrossRef Full Text | Google Scholar Cluster Multiple Labeling Technique and Critical Concentration Algorithm CrossRef Full Text | Google Scholar “Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift,” in Proceedings of the 32nd International Conference on Machine Learning Google Scholar Google Scholar Quantitative Characterization of the Microstructure and Transport Properties of Biopolymer Networks PubMed Abstract | CrossRef Full Text | Google Scholar (Algorithms and Computation in Mathematics) Google Scholar Linking Morphology of Porous Media to Their Macroscopic Permeability by Deep Learning CrossRef Full Text | Google Scholar Principles of Heat Transfer in Porous Media Google Scholar Determination of the Dzyaloshinskii-Moriya Interaction Using Pattern Recognition and Machine Learning CrossRef Full Text | Google Scholar Local Tortuosity Inhomogeneities in a Lithium Battery Composite Electrode CrossRef Full Text | Google Scholar Über kapillare Leitung des Wassers im Boden: Aufstieg Versickerung und Anwendung auf die Bewässerung Google Scholar Influence of the Electrolyte Salt Concentration on the Rate Capability of Ultra‐Thick NCM 622 Electrodes CrossRef Full Text | Google Scholar “A Simple Weight Decay Can Improve Generalization,” in Proceedings of the 4th International Conference on Neural Information Processing Systems Google Scholar The Lattice Boltzmann Method – Principles and Practice Google Scholar CrossRef Full Text | Google Scholar Lantuéjoul Geostatistical Simulation: Models and Algorithms Google Scholar Lantuéjoul Geodesic Methods in Quantitative Image Analysis CrossRef Full Text | Google Scholar Google Scholar “Efficient BackProp,” in Neural Networks: Tricks of the Trade CrossRef Full Text | Google Scholar Impact of Microstructure on the Effective Diffusivity in Random Packings of Hard Spheres CrossRef Full Text | Google Scholar The Kinetics of Precipitation from Supersaturated Solid Solutions CrossRef Full Text | Google Scholar Precise Algorithms to Compute Surface Correlation Functions of Two-phase Heterogeneous Media and Their Applications PubMed Abstract | CrossRef Full Text | Google Scholar Google Scholar Physical Models of Diffusion for Polymer Solutions CrossRef Full Text | Google Scholar Google Scholar Google Scholar A Linear Time Algorithm for Computing Exact Euclidean Distance Transforms of Binary Images in Arbitrary Dimensions CrossRef Full Text | Google Scholar A Time-Optimal Algorithm for the Estimation of Contact Distribution Functions of Random Closed Sets CrossRef Full Text | Google Scholar A Unified Simulation Framework for Spatial Stochastic Models CrossRef Full Text | Google Scholar The Cahn–Hilliard Equation: Recent Advances and Applications Google Scholar Stochastic Geometrical Modeling of Solid Oxide Cells Electrodes Validated on 3D Reconstructions CrossRef Full Text | Google Scholar Contradicting Geometrical Concepts in Pore Size Analysis Attained with Electron Microscopy and Mercury Intrusion CrossRef Full Text | Google Scholar Google Scholar Estimation of Geodesic Tortuosity and Constrictivity in Stationary Random Closed Sets CrossRef Full Text | Google Scholar On a Pluri-Gaussian Model for Three-phase Microstructures with Applications to 3D Image Data of Gas-Diffusion Electrodes CrossRef Full Text | Google Scholar Stochastic 3D Modeling of Complex Three-phase Microstructures in SOFC-Electrodes with Completely Connected Phases CrossRef Full Text | Google Scholar Quantifying the Influence of Microstructure on Effective Conductivity and Permeability: Virtual Materials Testing CrossRef Full Text | Google Scholar Google Scholar The Electrode Tortuosity Factor: Why the Conventional Tortuosity Factor Is Not Well Suited for Quantifying Transport in Porous Li-Ion Battery Electrodes and What to Use Instead CrossRef Full Text | Google Scholar Statistical Analysis of Microstructures in Materials Science Google Scholar 3D Images of Materials Structures: Processing and Analysis Google Scholar Stochastic 3D Microstructure Modeling of Anodes in Lithium-Ion Batteries with a Particular Focus on Local Heterogeneity CrossRef Full Text | Google Scholar “Generating Digital Twins of Mesoporous Silica by Graph-Based Stochastic Microstructure Modeling CrossRef Full Text | Google Scholar Data from: Large-scale Statistical Learning for Mass Transport Prediction in Porous Materials Using 90,000 Artificially Generated Microstructures [Data set] Zenodo CrossRef Full Text | Google Scholar “Parametric Microstructure Modeling of Compressed Cathode Materials for Li-Ion Batteries CrossRef Full Text | Google Scholar Prediction of Effective Properties of Porous Carbon Electrodes from a Parametric 3D Random Morphological Model CrossRef Full Text | Google Scholar On the Momentum Term in Gradient Descent Learning Algorithms PubMed Abstract | CrossRef Full Text | Google Scholar Predicting Permeability via Statistical Learning on Higher-Order Microstructural Information PubMed Abstract | CrossRef Full Text | Google Scholar Computational Screening of Diffusive Transport in Nanoplatelet-Filled Composites: Use of Graphene to Enhance Polymer Barrier Properties CrossRef Full Text | Google Scholar Computational High-Throughput Screening of Fluid Permeability in Heterogeneous Fiber Materials PubMed Abstract | CrossRef Full Text | Google Scholar Shape-dependent Effective Diffusivity in Packings of Hard Cubes and Cuboids Compared with Spheres and Ellipsoids PubMed Abstract | CrossRef Full Text | Google Scholar “Effective Diffusivity in Lattices of Impermeable Superballs CrossRef Full Text | Google Scholar Google Scholar Flow and Transport in Porous Media and Fractured Rock: From Classical Methods to Modern Approaches Google Scholar “Mechanical Behavior Predictions of Additively Manufactured Microstructures Using Functional Gaussian Process Surrogates CrossRef Full Text | Google Scholar “Measuring Intrinsic Volumes in Digital 3D Images,” in 13th International Conference Discrete Geometry for Computer Imagery Google Scholar Deep Learning in Neural Networks: An Overview PubMed Abstract | CrossRef Full Text | Google Scholar Google Scholar Image Analysis and Mathematical Morphology Google Scholar Morphological Image Analysis: Principles and Applications Google Scholar Srisutthiyakorn “Deep-learning Methods for Predicting Permeability from 2D/3D Binary-Segmented Images,” in 2016 SEG International Exposition and Annual Meeting CrossRef Full Text | Google Scholar Dropout: A Simple Way to Prevent Neural Networks from Overfitting Google Scholar Spatial Modeling of the 3D Morphology of Hybrid Polymer-ZnO Solar Cells CrossRef Full Text | Google Scholar Big Data for Microstructure‐property Relationships: A Case Study of Predicting Effective Conductivities CrossRef Full Text | Google Scholar Predicting Effective Conductivities Based on Geometric Microstructure Characteristics CrossRef Full Text | Google Scholar Driving Digital Rock towards Machine Learning: Predicting Permeability with Gradient Boosting and Deep Neural Networks CrossRef Full Text | Google Scholar Quantifying Tortuosity in Porous Li-Ion Battery Materials CrossRef Full Text | Google Scholar Tortuosity in Electrochemical Devices: A Review of Calculation Approaches CrossRef Full Text | Google Scholar Random Heterogeneous Materials - Microstructure and Macroscopic Properties Google Scholar Random Heterogeneous Media: Microstructure and Improved Bounds on Effective Properties CrossRef Full Text | Google Scholar The Relative Neighbourhood Graph of a Finite Planar Set CrossRef Full Text | Google Scholar Stochastic Modelling of 3D Fiber Structures Imaged with X-ray Microtomography CrossRef Full Text | Google Scholar Analyzing Machine Learning Models to Accelerate Generation of Fundamental Materials Insights CrossRef Full Text | Google Scholar van der Linden Machine Learning Framework for Analysis of Transport through Complex Networks in Porous PubMed Abstract | CrossRef Full Text | Google Scholar An interior Algorithm for Nonlinear Optimization that Combines Line Search and Trust Region Steps CrossRef Full Text | Google Scholar Modelling Mesoporous Alumina Microstructure with 3D Random Models of Platelets CrossRef Full Text | Google Scholar Prediction of Effective Diffusivity of Porous media Using Deep Learning Method Based on Sample Structure Information Self-Amplification CrossRef Full Text | Google Scholar Generation of Virtual Lithium-Ion Battery Electrode Microstructures Based on Spatial Stochastic Modeling CrossRef Full Text | Google Scholar Investigation of the Relationship between Morphology and Permeability for Open-Cell Foams Using Virtual Materials Testing CrossRef Full Text | Google Scholar Stochastic Modeling and Predictive Simulations for the Microstructure of Organic Semiconductor Films Processed with Different Spin Coating Velocities CrossRef Full Text | Google Scholar Physics and Mechanics of Random Media: From Morphology to Material Properties (Paris: Presses des MINES) Google Scholar Predicting Effective Diffusivity of Porous Media from Images by Deep Learning PubMed Abstract | CrossRef Full Text | Google Scholar Seeing Permeability from Images: Fast Prediction with Convolutional Neural Networks CrossRef Full Text | Google Scholar A Measure of General Functional Dependence between Two Continuous Variables CrossRef Full Text | Google Scholar On Pressure and Velocity Boundary Conditions for the Lattice Boltzmann BGK Model CrossRef Full Text | Google Scholar Neumann M and Schmidt V (2021) Large-Scale Statistical Learning for Mass Transport Prediction in Porous Materials Using 90,000 Artificially Generated Microstructures Received: 30 September 2021; Accepted: 27 October 2021;Published: 23 December 2021 Copyright © 2021 Prifling, Röding, Townsend, Neumann and Schmidt. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use distribution or reproduction in other forums is permitted provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited in accordance with accepted academic practice distribution or reproduction is permitted which does not comply with these terms *Correspondence: Benedikt Prifling, YmVuZWRpa3QucHJpZmxpbmdAdW5pLXVsbS5kZQ== †These authors have contributed equally to this work Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher 94% of researchers rate our articles as excellent or goodLearn more about the work of our research integrity team to safeguard the quality of each article we publish Our websites may use cookies to personalize and enhance your experience. By continuing without changing your cookie settings, you agree to this collection. For more information, please see our University Websites Privacy Notice. Ashley Roding reflects on her time at UConn Ashley Roding is graduating from UConn with a degree in pharmacy science – and it is the second undergraduate degree in her life. She discovered a passion for pharmacy while studying for her first degree in biomedical science, and decided to pursue another. She will remain at UConn for two more years and study for a Pharm.D. degree. What’s your major/field of study, and what drew you to it? Did you have a favorite professor or class? What activities were you involved in as a student? What are your plans after graduation/receiving your degree? After graduation, I will be at UConn for another two years to complete my Doctor of Pharmacy degree! My goal is to complete two years of residency after receiving my doctorate, and ultimately go into a career in clinical pediatric pharmacy. How has UConn prepared you for the next chapter in life? UConn has given me so many opportunities to grow as a person, not only through traditional classes but through all of the unique opportunities I’ve gotten to be a part of. For example, I created an outreach project for the pediatric track on Prescription Drug Abuse that I was able to present at Coventry High School with a team of other students. Experiences like this gave me the chance to build and expand on skills I’ve learned in classes. Don’t be afraid to put yourself out there! During my first year at UConn, I explored many different organizations at the involvement fair, and have since then grown into leadership roles I couldn’t have imagined having. If I hadn’t put myself out there, I would not have grown and developed into the person I am today. Volume 9 - 2022 | https://doi.org/10.3389/fmats.2022.956839 Small-angle X-ray scattering (SAXS) is a useful technique for nanoscale structural characterization of materials structural and spatial information is indirectly obtained from the scattering intensity in the spectral domain characterizing the structure requires solving the inverse problem of finding a plausible structure model that corresponds to the measured scattering intensity Both the choice of structure model and the computational workload of parameter estimation are bottlenecks in this process we develop a framework for analysis of SAXS data from disordered materials The materials are modeled using Gaussian Random Fields (GRFs) where a third phase is added at the interface between the two other phases Fourier transform-based numerical methods for both structure generation and SAXS simulation We demonstrate that length scales and volume fractions can be predicted with good accuracy using our machine learning-based framework The parameter prediction executes virtually instantaneously and hence the computational burden of conventional model fitting can be avoided To characterize detailed random porous structures high-resolution 3D imaging techniques e.g. micro/nano X-ray computed tomography (X-ray CT) focused ion beam scanning electron microscopy (FIB-SEM) and transmission electron microscopy tomography (TEMT) can provide high quality information on morphological features imaging techniques are frequently time-consuming and require advanced sample preparation methods or sample environments the attainable contrast is strongly sample-dependent and can be prohibitively low fitting of non-analytical models can be computationally prohibitive where the third phase is an intermediate layer residing by the interface between pore and solid We generate a large number of virtual microstructures for a number of cases We further develop a Fourier transform-based numerical method for simulating realistic SAXS data Both the structure generation and SAXS simulation methods are heavily optimized and implemented on GPU with a combined execution time in the order of 1 s Using the simulated SAXS data as input and the known generation parameters as target output in a machine learning framework we demonstrate that length scales and volume fractions can be predicted with good accuracy Our framework is a proof of concept that can be applied to other types of disordered materials as well in particular other Gaussian random field-based models with different covariance structures Since the scattering arises from the interaction between X-rays and the electron clouds of the atoms theoretically the scattering intensity can be written as The SAXS curve consists of measured values of I(q) for a large number of typically equidistant q values in the range qmin ≤ q ≤ qmax Assume that a virtual electron density ρ is simulated on a periodic cubic domain with resolution N3 and voxel size Δx Then the fast Fourier transform (FFT) can be used to obtain the discrete counterpart to the scattering intensity It is also defined on a periodic cubic domain with resolution N3 for all qijk=qx(i),qy(j),qz(k) such that An orientation-averaged scattering intensity I(q) (an “intensity data reduction”) can then be computed as for qijk = |qijk| Δq = 2π/(NΔx) (the grid resolution in q space) with wq0=0 and wq,0 chosen so that the sum of the weights is 1 wq describes a normal distribution in the radial direction with mean q and standard deviation Δq/2 all weights rescaled with 1/qijk2 to compensate for the fact that the number of grid points increases in proportion to 4πqijk2 (the area of a spherical shell with radius qijk) In practice, because of the symmetries in q space, the N3 array I(q) can be folded into a (N/2 + 1)3 array which substantially reduces the computation time (for the steps corresponding to Eqs 3, 4) The simulation is implemented on GPU in Matlab (Mathworks In the original model by Cahn and Hilliard (Cahn and Hilliard, 1958) GRFs arose as a solution to a spinodal decomposition model described as a superposition of cosine waves some wave vectors qm and random phase offsets ηm 0 ≤ ηm < 2π The wave vectors follow some probability distribution Γq; if it is radially symmetric and only a function of |q| A disadvantage of constructing a GRF like this is that ψ(x) is not a periodic function unless the wave vectors are constrained to axis-aligned directions and certain magnitudes Therefore, we instead use a method based on the Fast Fourier Transform (FFT) (Lang and Potthoff, 2011). A GRF can generally be described by a mean value and a covariance function (Liu et al., 2019). The spectral density of this covariance function actually equals Γq (Teubner, 1991) Generating a GRF in a cubic domain with resolution N3 is performed as follows Gaussian noise is generated in the spatial domain it is Fourier transformed and multiplied by the square root of the spectral density of the target covariance function yielding a GRF ψ(x) with the specified covariance function where W is N(0,1)-distributed and independent for all x a special case of a spectral density used before (Matérn, 1986; Lang and Potthoff, 2011), also in models for materials microstructures (Röding et al., 2020; Prifling et al., 2021) (note that the spectral density is not normalized hence not a probability distribution; this only results in a linear scaling of the GRF Because FW and Γ are both symmetric The parameter a has dimension length and we refer to it as a scaling parameter to which the length scale is approximately proportional It is important to note that the length scale is determined not only by a but by the entire functional form of Γ Microstructures are generated from the GRFs in the following manner a binary function is obtained by thresholding for T such that p (ψ(x) ≤ T) = ϵ ψ′ is smoothed with a 3D Gaussian filter (accounting for periodicity; σ = 2 voxels 0 ≤ ψ″ ≤ 1 Two-phase electron densities can now be defined by for T such that p (ψ″(x) ≤ T) = ϵ Three-phase electron densities can be defined similarly by for T1 such that p (ψ″(x) ≤ T1) = ϵ(1 − ν) and T2 such that p (T1 < ψ″(x) ≤ T2) = ϵν) (and also p (ψ″(x) > T2) = 1 − ϵ)) Note that the ‘intermediate’ values of ψ″ (not equal to 0 or 1) will be concentrated near what will be the pore-solid interface which is the reason for defining it this way the intermediate layer will be adjacent to both pore and solid ρ(x) would be obtained by thresholding ψ(x) directly parts of the intermediate layer phase might end up in contact only with pore or only with solid in all directions It is also worth pointing out that by using the 3D Gaussian filter also in the two-phase model the two- and three-phase models are seamlessly integrated into the same framework and the two-phase model is a special case of the three-phase model The generation procedure is illustrated with microstructures generated on a grid of size 5123 with voxel size Δx = 0.5 nm (therefore the parameter a also has unit nm). In Figure 1, using a single 2D slice of a three-phase model with a = 4 nm, ϵ = 0.60, and ν = 0.50. In Figure 2 a 3D visualization of the same structure is shown Illustration of a single 2D slice of a three-phase model with a =4 nm the GRF is shown (arbitrary intensity scale) obtained from (A) by thresholding at the quantile ϵ obtained from (B) by smoothing with a 3D Gaussian filter obtained from (C) by thresholding at the quantiles ϵ(1− ν) and ϵ FIGURE 2. Illustration of a three-phase model with a =4 nm, ϵ =0.60, and ν =0.50. For an illustration of a single 2D slice from this model, see Figure 1 For developing a machine learning-based model for prediction of microstructural parameters a large number of microstructures are generated on a grid of size 5123 with voxel size Δx = 0.5 nm (we reiterate that therefore the parameter a also has unit nm whereas ϵ and ν are dimensionless quantities) We reiterate that two-phase microstructures are generated using ρpore = 0 and ρsolid = 1; varying ρsolid is unnecessary because SAXS data can always be rescaled Three-phase microstructures are generated using the same values and additionally ρlayer = 0.65 where Δρ is the electron density difference between the two phases whether the structure has low or high porosity is information that has to be supplied by the user and unless ρlayer = 1/2 there is no exact mirror image we treat two- and three-phase structures consistently in this respect 219 (524,288) microstructures are generated for the training dataset The scattering intensity is simulated for 500 q values equidistant between qmin ≈ 0.04 nm−1 and qmax ≈ 3.00 nm−1; these values are taken from an in-house experimental setup using an Anton Paar SAXSpoint 2.0 (Anton Paar Austria) and cover a normal SAXS probing range without losing the generality the average execution time for microstructure generation and simulation of SAXS data combined is approximately 1 s A Poisson model would imply that the variance of the noise is σ2(q) = I(q); however considering the intensity scales resulting from using I0 = 1 this model assumption does not produce realistic noise levels we use a lognormal noise model with mean I(q) and σ2(q) = αI(q) where a value of α is sampled from a log-uniform distribution in [102 The lognormal distribution only produces positive values whereas the commonly suggested normal approximation can produce physically implausible Note that whereas we account for measurement noise we do not account for the finite instrument resolution which would yield a slight blurring of the SAXS curves Further, analogously to Gommes (2018) (Gommes, 2018) dividing by a total intensity approximated by where Q̃ is a discrete approximation of the Porod invariant. Note that this is a summation of the intensities weighted by spherical shells with thickness δq, the distance between consecutive values in the vector of q values. Effectively, 4πQ̃ equals the total intensity in a spherical shell with inner and outer radii qmin and qmax. In Figure 3 examples of simulated SAXS curves are shown Note that for some parameter values (specifically for some scaling parameters) the low-q plateau is very long and carries very little information for parameter prediction when selecting a q range appropriate for the entire data set it is unavoidable that some SAXS curves exhibit a long plateau a is varied for ϵ =0.3 and ν =0 (two-phase model) ϵ is varied for a =2.5 nm and ν =0 (two-phase model) ν is varied for a =2.5 nm and ϵ =0.3 (three-phase model) measurement noise has not been added to these curves The inputs of the training set are first preprocessed by transforming to logarithmic scale they are standardized by computing the mean and variance for each dimension separately on the training set and then rescaling to zero mean and unit variance the same rescaling (using the mean and variance of the training set) is applied to the validation and test sets (and also to new data once the prediction model is finalized and used) the trees of the XGBoost model are optimized with respect to mean squared error (MSE) loss The choice of which hyperparameters to study and their ranges are selected after an initial investigation this is performed only for the ϵ parameters (ϵ and ν are the hardest to predict and of those only ϵ is present in all datasets) The results led to the following hyperparameters being used for all cases: learning_rate = 0.005 training is performed with these values and n_estimators = 50,000 10 training runs are performed and the best-performing model is selected We also use an early stopping rule that finalizes the training if no improvement is found for the last 1,000 added trees and the best-performing model (best value of n_estimators) is selected the values of n_estimators in the final models vary from 6,060 to 48,841 The average execution time is approximately 3 h with a brief explanation of their meaning and range of their values The results for the final selected XGBoost models is shown in Table 2 we also use the more intuitive mean absolute percentage error (MAPE) loss Error measures for the prediction of the parameters where MSE and MAPE (in %) is given for the training which is determined by the distribution of inputs and outputs in the training set and the prediction model itself It is also worth pointing out that if the prediction model would have been trained to predict the porosity on the low-porosity and the high-porosity data jointly the reported accuracy would be substantially lower Scatter plots showing prediction results on the test set for both two-phase datasets predictions of a and ϵ are shown for the two-phase low porosity structures predictions of a and ϵ are shown for the two-phase high porosity structures Scatter plots showing prediction results on the test set for both three-phase datasets and ν are shown for the three-phase low porosity structures and ν are shown for the three-phase high porosity structures it is important to note that the structures are random and not uniquely defined by the set of parameter values used to generate them; each set of parameter values can yield a very large number of different structures that in turn yield an equal number of different SAXS curves a SAXS curve cannot be uniquely mapped to a set of parameter values even in the absence of measurement noise; their relationship is inherently random It follows that the prediction loss is due to a combination of the randomness of the structures and the randomness induced by the added measurement noise there is in practice a lower bound on the attainable accuracy This effect is essentially a result of the limited resolution and field of view of the simulated data and not as such a fundamental limitation of SAXS It is worth noting that we investigate two other techniques for regression The first is also based on XGBoost but utilizing chained regression This means that the different outputs are predicted sequentially such that the predictions of the first are used as input for prediction of the second and the predictions of the first and second are used as inputs for prediction of the third Also we investigate fully-connected artificial neural networks An initial investigation suggests that neither of these two attempts yield better results than the ‘plain’ XGBoost approach presented showing the simulated 1D SAXS curve from both the true structure and a reconstructed structure using the predicted parameter values (in this case â=3.982 nm and ν̂=0.195) as well as representative slices from the true and reconstructed structures the reconstructed SAXS curve reasonably well reproduces that of the true structure due to the simulated SAXS curves being an average of very few values of I(q) for low q the random fluctuations between different structures will be larger in that range Histogram of estimated values for 500 simulated SAXS curves for a = 4 nm The true values are also indicated (vertical black lines) Results for a single structure in the case study the simulated SAXS curve for a =4 nm and a (noiseless) SAXS curve of a structure generated using the predicted parameters â=3.982 nm and ν̂=0.195 (red) are shown single slices from the true structure and the reconstructed structure are shown We have implemented a machine learning-based approach to fast estimation of microstructural parameters from SAXS data The microstructure model is based on a periodic Gaussian random field with variable length scale which is processed and thresholded to yield two-phase (pore and solid) and three-phase (pore with all phases having different electron densities We also develop a Fourier transform-based method to simulate SAXS data Both microstructure generation and SAXS simulation are implemented on the GPU and very fast We demonstrate that by performing regression using XGBoost a decision tree-based machine learning framework the parameters of the models can be predicted with good accuracy Given that artificial neural networks did not perform better than XGBoost and given that there is no time dependence or translational invariance in the data to further exploit it is unlikely that more advanced architectures such as recurrent or convolutional neural networks would perform better the parameter prediction executes virtually instantaneously Hence the computational burden of conventional model fitting can be avoided enabling for the SAXS practitioner to efficiently analyze many measurements We observed positive and negative bias in the predictions observed near the lower and upper bound of the simulated parameter ranges This bias could be reduced by using a wider range of parameters (where possible) for the training set while maintaining the ranges for the validation and test sets the performance of the prediction will be assessed in a smaller parameter space which should then be considered the domain of applicability Although the microstructure models herein are aimed at mimicking a certain type of morphology and certain ranges of the parameters similar models can be expected to perform well for other types of microstructures (i.e. The only requirement is that the microstructure model is efficiently implemented so that a large and that the corresponding SAXS curves are sufficiently informative regarding the parameters to be predicted Although the approach is evaluated on a specific type of morphology it is a proof of concept that can be used for other types of materials both with regard to spatial structure and electron density values and also for other experimental parameters such as other q value ranges generalizing this investigation to multiple classes of Gaussian random field-based models would be an interesting prospect for further work this proof of concept illustrates the usefulness not only of the machine learning-based approached but also of the efficient GPU-accelerated scheme for simulating the materials structures and the corresponding SAXS data and the new three-phase model all the data and codes used herein are publicly available to facilitate further development in this field The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: DOI:10.5281/zenodo.5948941 MR developed the microstructure generation and SAXS simulation methods together with SY and MB developed the machine learning methods All authors contributed to designing the study and to writing the manuscript MR acknowledges the financial support of the Swedish Research Council for Sustainable Development (grant number 2019-01295) SY acknowledges the financial support of the Swedish Research Council (grant number 2018-06378) Classifying and analyzing small-angle scattering data using weighted k nearest neighbors machine learning techniques CrossRef Full Text | Google Scholar Prediction of diffusive transport through polymer films from characteristics of the pore geometry CrossRef Full Text | Google Scholar Random search for hyper-parameter optimization Google Scholar Scattering properties of the leveled-wave model of random morphologies PubMed Abstract | CrossRef Full Text | Google Scholar Small-angle X-ray scattering on biological macromolecules and nanocomposites in solution doi:10.1146/annurev-physchem-040412-110132 PubMed Abstract | CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Measurement of the Gaussian curvature of the surfactant film in an isometric bicontinuous one-phase microemulsion PubMed Abstract | CrossRef Full Text | Google Scholar “A scalable tree boosting system,” in Proceedings of the 22nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining Google Scholar PubMed Abstract | CrossRef Full Text | Google Scholar D’hollander Modeling the morphology and mechanical behavior of shape memory polyurethanes based on solid-state NMR and synchrotron SAXS/WAXD CrossRef Full Text | Google Scholar Small angle scattering data analysis assisted by machine learning methods CrossRef Full Text | Google Scholar Stochastic 3D modeling of the microstructure of lithium-ion battery anodes via Gaussian random fields on the sphere CrossRef Full Text | Google Scholar Machine learning methods for X-ray scattering data analysis from biomacromolecular solutions CrossRef Full Text | Google Scholar Topology-generating interfacial pattern formation during liquid metal dealloying PubMed Abstract | CrossRef Full Text | Google Scholar Stochastic analysis of capillary condensation in disordered mesopores PubMed Abstract | CrossRef Full Text | Google Scholar Structure development of resorcinol-formaldehyde gels: Microphase separation or colloid aggregation PubMed Abstract | CrossRef Full Text | Google Scholar Stochastic models of disordered mesoporous materials for small-angle scattering analysis and more CrossRef Full Text | Google Scholar Three-dimensional reconstruction of liquid phases in disordered mesopores using in situ small-angle scattering CrossRef Full Text | Google Scholar Model reconstruction from small-angle X-ray scattering data using deep learning methods PubMed Abstract | CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Morphological characterization of bicontinuous phase-separated polymer blends and one-phase microemulsions CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Small angle X-ray scattering for nanoparticle research PubMed Abstract | CrossRef Full Text | Google Scholar Advances in Gaussian random field generation: A review CrossRef Full Text | Google Scholar Three-dimensional bicontinuous nanoporous materials by vapor phase dealloying PubMed Abstract | CrossRef Full Text | Google Scholar Google Scholar present and future of lithium-ion batteries: Can new technologies open up new horizons?,” in Lithium-ion batteries CrossRef Full Text | Google Scholar Large-scale statistical learning for mass transport prediction in porous materials using 90 000 artificially generated microstructures CrossRef Full Text | Google Scholar Versatility and robustness of Gaussian random fields for modelling random media CrossRef Full Text | Google Scholar Morphology and thermal conductivity of model organic aerogels CrossRef Full Text | Google Scholar Predicting permeability via statistical learning on higher-order microstructural information PubMed Abstract | CrossRef Full Text | Google Scholar Advances in the development of sol-gel materials combining small-angle X-ray scattering (SAXS) and machine learning (ML) CrossRef Full Text | Google Scholar Simulation of small-angle scattering curves by numerical Fourier transformation CrossRef Full Text | Google Scholar Quantitative evaluation of statistical errors in small-angle X-ray scattering measurements PubMed Abstract | CrossRef Full Text | Google Scholar Tabular data: Deep learning is not all you need CrossRef Full Text | Google Scholar Small-angle X-ray scattering intensity of multiscale models of spheres CrossRef Full Text | Google Scholar Identifying domains of applicability of machine learning models for materials science PubMed Abstract | CrossRef Full Text | Google Scholar PubMed Abstract | CrossRef Full Text | Google Scholar Level surfaces of Gaussian random fields and microemulsions CrossRef Full Text | Google Scholar “Machine learning-assisted analysis of small angle X-ray scattering,” in Proceedings of the 2021 Swedish Workshop on Data Science PubMed Abstract | CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Small-angle X-ray scattering of nanoporous materials CrossRef Full Text | Google Scholar Nanocavitation in carbon black filled styrene–butadiene rubber under tension detected by real time small angle X-ray scattering CrossRef Full Text | Google Scholar Borg M and Rönnols J (2022) Machine learning-accelerated small-angle X-ray scattering analysis of disordered two- and three-phase materials Received: 30 May 2022; Accepted: 30 August 2022;Published: 27 September 2022 Copyright © 2022 Röding, Tomaszewski, Yu, Borg and Rönnols. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use *Correspondence: Magnus Röding, bWFnbnVzLnJvZGluZ0ByaS5zZQ== The unloved waterway that winds its way to the Thames has a staunch defender Paul Powlesland has spent five years calling out polluters and keeping footpaths open – sometimes with the law An ancient oak stretches huge branches across the glinting water and London’s third biggest river looks as graceful as it did two centuries ago Then my walk with Paul Powlesland on the banks of the River Roding collides with contemporary Britain sending a steady trickle of sewage into the main channel failing to provide residents with any greenery or way to access the river heaps of fly-tipped rubbish and the recent relics of camps created by desperate people without homes: mouldering mattresses found other local people wanting to clean up their river ‘It has pretty much all of the water quality issues it’s possible for a river to have’ … Powlesland on his rowing boat – his oars had just been stolen Photograph: Jill Mead/The GuardianThere is a lot to do the companies releasing pollutants or sewage are being held to account and developers are being cajoled into providing riverside homes that better serve the river and its neighbours looking as if he has stepped from the set of Buck Rogers just stepped from an equally far-fetched scene: his off-grid boat beside moorings he built from scaffolding and which is equipped with a wood-fired sauna and hot tub and outdoor barbecue area to tempt other boat-dwellers to join him as resident “river guardians” From Stansted to the Thames … the river between Barking and Ilford Photograph: Jill Mead/The GuardianThe tide is flowing swiftly upriver on a bright morning as we begin our walk north up steps on to a footpath beside the river “These steps are a prime example of why you need local nature guardians and also how power works in this country,” says Powlesland “This is a public footpath; it’s been used for over 30 years But a developer wanted to use this area for materials storage so they just boarded it up and shut off the path I came in with a sledgehammer and smashed it open.” I assume Powlesland is using a legal metaphor No: the barrister hacked down the wooden hoardings “I’m legally entitled to do it because they are blocking a public highway,” he explains “Knowing the law helped me do the direct action because I wasn’t committing an offence – they were and I was just correcting it on the ground.” “Eventually they rebuilt it so strong that I couldn’t smash it any more so then I had to start local campaigning The path here was shut for a year and a half And this route allows the ordinary people of Barking to access Tesco from the town centre without walking along really busy roads.” After Powlesland campaigned via the River Roding Trust the developer eventually reopened the path He fell in love with nature in his 20s (various festivals helped alter his mindset he swims in the Roding – and tree-planting are combined in his river guardianship with the trust organising work-days where up to 30 local volunteers build benches and perform Herculean litter-picking tasks But Powlesland also undertakes guerrilla planting “There’s a willow I planted.” Along the bank dozens of young willows are growing rapidly “Imagine you’re having a Sunday out in Barking in February and you see a man in a bright pink jumpsuit poling a kind of Venetian gondola a giant muddy raft full of willow saplings I just pop them in every bit of mud you see.” Next to the A406 North Circular Road Powlesland checks on a sapling near his narrowboat home Photograph: Jill Mead/The GuardianPowlesland held a willow-planting birthday party three years ago where the river bends prettily around a large reedbed It’s just incredible!” A three-year-old willow’s trunk is already as thick as a Jack Grealish calf “I’ve never climbed a tree I’ve planted before,” says Powlesland “There’s few greater things in life than being able to climb a tree you planted.” Willow is a “miracle tree” and free – it grows from cuttings so Powlesland chops off a few young branches and turns them into new trees handy in this summer’s drought when Powlesland wheelbarrowed water around to sustain other species of newly planted trees A grey heron Photograph: Jill Mead/The Guardian“I’ve come to see the importance of cycling between the micro and the macro,” says Powlesland “If I just work in Barking it’s irrelevant because these trees will get drowned by climate change It’s a tsunami of single-use packaging and it feels like trying to wipe off the overflow from the bath rather than turning off the tap and for packaging we really need to turn off the tap the local helps keep your sanity and keep you motivated.” Powlesland deployed his barrister skills in the high court to try to save a black poplar south London but his opponents won the case and the tree was removed I had a load of black poplars to plant down here I dedicated them all to the barrister and the judge whose actions had led to the destruction of the Wandsworth black poplar.” Free daily newsletterOur morning email breaks down the key stories of the day telling you what’s happening and why it matters Powlesland speaking at an Extinction Rebellion protest against Thames Water pollution in February Photograph: Jessica Girvan/AlamyThe species may be rare but “Barking now has one of the best collections of black poplars in the country,” grins Powlesland “There are about 100 of them but no one knows about it yet because they are still saplings We got absolutely no permission from anyone The first thing they’ll know about it is when Barking is transformed – ‘Where do all these massive trees come from?’” Won’t some jobsworth order them to be removed “Barking and Dagenham council have been very supportive of the River Roding Trust and gave us some money to plant trees as well.” Our stroll continues along an unofficial riverside path The trust has received £50,000 of funding to make this an official path but Redbridge council Opening up this precious urban green space to people is crucial to revive the river The closed riverbank is currently filled with harrowing rubbish-strewn camps left by homeless people “People don’t have the headspace to think about nature’s needs and nature gets trashed This area is really lacking in nature and green space inaccessible stream was too difficult to monitor So he cut another unofficial path and monitors it himself “One of the richest companies in the country can’t or won’t do it and no one in authority says to them: ‘You’re not a criminal gang; if there’s a risk of you committing a repeated criminal offence at least check it’s not routinely happening.’” A Thames Water spokesperson says blockages in the area’s complex and old “foul” sewage system is causing sewage to “cross over” into the surface network “Thames Water’s engineering department is currently designing a solution that will make an alteration to the network at the point of discharge and divert polluted surface flows into the foul system This will prevent blockages in future causing pollution to the Alders Brook while still retaining hydraulic capacity of the network to absorb surface water and prevent homes flooding.” Read moreThe joy of becoming a nature guardian We don’t need to wait for someone else to change the law The current legal framework is not conducive to nature protection but it’s better to have people on the ground connecting with rivers than passing a fancy law which no one is going to uphold.” And nature guardianship is not just a duty “We know nature benefits from it but the more you put into a place the more you grow to love it and the more you get out of it.” For all the painful reality of saving an abused river too: discovering sand martins from sub-Saharan Africa nesting in old drainage pipes at the river edge watching kingfishers flash past in their hunt for fish savouring the fruiting of apple and peach trees inadvertently planted by motorists hurling cores from their cars on the North Circular; all the amazing signs of the natural world’s indefatigable lust for life that matches his own Come along on our journey and stay up to date the days and weeks leading up to the release event among supporters journalists and politicians were turbulent we worked on the prototype the night before the presentation Our team had discovered an error in the cabling during the final check then as now jointly responsible for vehicle development carefully drove the Sion down a ramp from the stage to the visitors who could not wait to see the prototype of the Sion at close range In the last three years we have completed thousands of test drives in Germany a total of about 15,000 kilometers landed on the digital speedometer We are certain: without these test drives we would never have been able to convince so many people of our concept But just as Sono Motors no longer consists of 15 employees who share a room the size of a single apartment these vehicles no longer represent the state of the art in development technology We have not only professionalized and developed ourselves as a company we have also raised the development of the Sion to a level that is ever closer to the final production version of the vehicle we think it is about time to send our first-generation prototypes into well-deserved retirement To be able to start building the new vehicles the manufacturer needs the current status of the design data of the Sion The transfer of the CAD data to our partner Roding was an important milestone for us which we were able to complete in June just as planned “CAD” is the English abbreviation for “computer aided design” This is because the design process of a vehicle includes the geometric modeling of the individual components as well as the calculation and simulation the manufacturer is responsible for the construction of the body shell and ensures that the vehicles are assembled as planned by our development team the starting signal for the construction of the new prototypes After careful consideration, our development team has decided to work with an old acquaintance: The Roding Automobile GmbH Already our first two prototypes have been built in 2017 in close cooperation with Roding – experts for lightweight construction and e-mobility Roding has been responsible for the maintenance and repair of the prototypes If we had stressed the vehicles a bit too much during the numerous test drives The local proximity was and is also very helpful for our future cooperation nobody knows more about the first generation prototypes than our colleagues from Roding They have been enjoying a proven expertise in the field of prototype production for over ten years many things still are taken care of manually and under the watchful eyes of co-founder and managing director Ferdinand Heindlmeier “At Roding we specialise in the construction of prototypes We were aware that we would not necessarily play a major role in the series development of the Sion” “The fact that Sono has now approached us again The construction of the prototypes in 2017 was an exciting and ambitious project that we enjoyed very much!” “I am particularly looking forward to seeing the vehicle live” but having the new vehicle within reach in our hall will certainly be a unique experience for everyone involved” Roding will slightly adjust the CAD model to prepare it for the appropriate manufacturing method During this process all necessary tools will be designed and manufactured specific parts are produced while our colleagues purchase the required vehicle components from our suppliers This is followed by the production of the first body shells Previously we had calculated that all these steps would be completed by the end of September as part of the program relaunch and taking into account the effects of the global corona pandemic This has delayed the prototype launch from the end of September to which can best be explained by why we develop prototypes at all and have them produced by an external company Philipp is responsible for the management of prototype development “A prototype is always a kind of snapshot of the development status” “Its main purpose is to show the progress of the development But the development team also learns a lot during the production process for example about the integration of the components and how well they already work together” In our conversation with vehicle developer Max we could already read that this type of “test run” involves all those areas that also play a central role in series production – from purchasing and logistics “There are two central reasons why we rely on an external manufacturer like Roding for the production of the prototypes” we save internal capacities which we use for further preparing the series production this is also one of the reasons why we have two vehicles manufactured instead of four by working with the external manufacturer we also learn a lot about our internal coordination processes which we can then optimize for later cooperation with the series manufacturer.” one of the most important questions remains: Where are the biggest differences between the old and the new generation of prototypes three years have passed since we presented our development status in the form of a vehicle most core components come from our series suppliers” and the complete chassis will already be installed in the SVC2 So the wheelbase and vehicle geometry is the same as they will be later in the series production this allows us to validate our simulations under real conditions our supporters can get a new impression of what it will feel like to drive the final vehicle The focus is clearly on a new look and feel” Over many years, we have taken our prototype vehicles very much to heart. Nevertheless, we can't wait for the new prototypes to arrive. Even though it will take a little more time than expected, we are very happy to have found the right partner to get the new prototypes on the road. We’ll keep you up-to-date with our newsletter and RSS feeds, and make sure that you always have the latest information when it comes to the Sion and Sono Motors. Press releases are provided by companies as is and have not been edited or checked for accuracy Any queries should be directed to the company issuing the release More Info: https://www.crown.com/en-us/about-us.html HOUSTON (May 1, 2025) – Mitsubishi Logisnext Americas one of the world’s leading manufacturers and providers of material handling today announced the completion of a major expansion at its Houston manufacturing campus with the 73,500-square-foot electrification fabrication facility This milestone marks a significant step in the company’s strategic initiative to grow its footprint in the rapidly expanding electric market The new facility is designed to meet the rising demand for Mitsubishi Logisnext Americas’ Electric Class I and Class II products It is also designed to significantly enhance production capacity and lower manufacturing costs to better serve evolving customer needs “This expansion is a reflection of our continued commitment to innovation and growth,” said Berry Mansfield President of Mitsubishi Logisnext Americas “It took a shared vision to bring this expansion to life This facility not only supports our growth in electrification but also creates a more engaging environment for our employees We’re excited to start production this year.” Construction began with a groundbreaking in August 2023 and was completed in spring 2024 including powder painting and robotic welding equipment culminating in the completion of the new electrification building in 2025 The development aligns with Mitsubishi Logisnext Americas’ long-term strategic goal to grow its warehouse product market share With electric-powered solutions continuing to outpace internal combustion (IC) vehicles in the market this facility strengthens the company’s position for sustained momentum Mitsubishi Logisnext Americas has invested nearly $20 million in the new facility to respond to increasing demand while improving lead times The new facility will help all parties involved by streamlining operations and boosting overall manufacturing efficiency To learn more about Mitsubishi Logisnext Americas and its network of dealers, please visit www.logisnextamericas.com a leading provider of intelligent warehouse automation solutions has appointed Cortney Hunt as Chief Operating Officer Hunt brings over 25 years of leadership in the strategy and implementation of warehouse technologies as an end user ensuring flawless delivery of the company's innovative automation solutions to customers Hunt served in senior leadership roles during his tenure at McLane Company Inc. most recently as Vice President of DC Network and Facilities he has consistently enhanced operational efficiency through the strategic implementation of automation and transportation technologies while building high-performing technical teams to support complex distribution center operations “Adding Cortney to our leadership team brings a genuine customer perspective to our solutions and implementation helping us deliver a better product” said Trew CEO “Cortney will add vast end user experiences to our team and lead driving operational improvements and helping ensure efficiency and alignment across the organization "I am excited to join Trew and further strengthen our position as a trusted partner delivering innovative automation solutions with the highest life-time value to our customers,” said Trew’s new COO “I am very grateful for my time at McLane Company Inc. and I look forward to helping them and our clients thrive.” Hunt holds both bachelor’s and master’s degrees in engineering from the University of Oklahoma He is a certified Lean Six Sigma Black Belt and former member of the University of Mary Hardin-Baylor College of Business Advisory Board makers of a Warehouse Management System (WMS) and AI-Powered Slotting Solution The acquisition will provide a strong technical foundation that enables Fidus to expand beyond equipment control into broader warehouse management accelerating progress to create the industry’s first truly open-architecture automation suite that is poised to revolutionize the future of warehouse automation Fidus will gain key assets from Fulfilld—a WMS system and an AI-powered slotting solution—that accelerate progress toward Fidus’ own open-architecture solutions that aim to provide customers true ownership and long-term control over their warehouse automation technology several key engineers from Fulfilld will join the Fidus team “Our core philosophy at Fidus has always been that technology should adapt to how people work not force people and businesses to adapt to the technology,” remarks Aarron Hale “and now with the integration of sophisticated core assets from Fulfilld we are able to take meaningful steps to create a comprehensive approach that ultimately frees businesses from the limitations of closed vendor-controlled systems across their entire warehouse operation We are excited to welcome our new team members from Fulfilld and plan to make significant strides towards an unrivaled open-architecture Enterprise Control Platform.” Fidus Global is committed to building the industry’s first truly open-architecture Enterprise Control Platform (ECP)—a comprehensive automation suite that will give operators control over every aspect of their material handling ecosystem Fidus Global is a warehouse software solutions firm and a full-service controls engineering firm with a diverse range of expertise that spans across retail Fidus provides clients with exceptional solutions in industrial automation with a team composed of seasoned engineers from leading corporations including Amazon Through the launch of Pontem (Best IT Innovation Award 2025 the industry’s first open architecture Warehouse Control System (WCS) that puts the customer back in control with both flexibility and scalability Fidus Global is reshaping the future of warehouse automation Fidusglobal.com systems and services for supply chain automation technology announced today important changes to the company’s leadership structure to build on an impressive growth history and to continue increasing market share will take on newly created roles on the company’s leadership team to better align the organization around demand generation and operational execution Johnson and Bruner have built Hy-Tek into an industry leader and successfully led multiple acquisitions “We are excited to take on these new roles and contribute to helping accelerate our growth by focusing on the customer and working with a talented group of people The future at Hy-Tek could not be brighter,” said Johnson “The growth of Hy-Tek has been amazing and it has been great to be a part of that We are looking forward to continuing to help shape the future by focusing on mission critical projects while mentoring our future leaders We will continue to push ourselves to operate at an ever-increasing level of high performance.” Using a strategy of organic growth and targeted acquisition Johnson and Bruner have been cornerstones in the exceptional growth of Hy-Tek and have led the Hy-Tek expansion into new services and vertical markets while successfully integrating multiple companies into the market leader Hy-Tek is today Johnson will now become Chief Customer Relations Officer for Hy-Tek Intralogistics Integrated Systems Division for Hy-Tek Material Handling Donnie will focus on strengthening relationships with customers while also mentoring the next generation of sales leaders at Hy-Tek to provide the best solutions for material handing needs Bruner will become Chief Operations Advisor after serving most recently as the Hy-Tek Intralogistics Chief Operating Officer troubleshooting and continuous improvement “Donnie and Mark have been the driving force in our business for many years and are recognized across the industry as leaders partners and friends,” said Hy-Tek Intralogistics CEO Kevin Viravec they will help us strengthen our enterprise account strategy and reach new levels of growth and strategic alignment with our partners We are excited to have them continuing with us on this journey.” Johnson will hand off systems leadership responsibility to Billy Carter Bruner will pass day-to-day operational and execution leadership to Brian Craft who has been promoted to SVP of Operations and will assume responsibility for engineering Johnson and Bruner will immediately begin working with the new senior leadership group on detailed plans to support their respective teams well into the future These moves are expected to increase Hy-Tek market share and customer satisfaction in the software and services sectors of the industry and will utilize cutting-edge technologies to match existing market leading sales design and engineering to produce the most effective one-stop shop for all The two companies are also gearing up for the launch of their first joint product “We are excited about the partnership with GreyOrange It allows enVista to have a single warehouse orchestration system provider to complement our automation and robotic systems integration business GreyOrange has the only warehouse orchestration platform in the market that optimizes independent robotic multi-agents within a facility regardless of the manufacturer,” said Jim Barnes “GreyOrange’s flexibility of integration supports one of enVista’s key differentiators which is to design and integrate automated facilities with any variety of autonomous robots This partnership further enables enVista to provide a unique level of innovation and creativity within our clients’ facilities facility optimization is about balancing the art of possibility with data science to create expansive operational excellence.” GreyOrange’s AI-powered software has enabled dozens of major retailers and logistics providers to embrace automation and ensure every aspect of the fulfillment process runs seamlessly inventory and people in warehouses and retail stores “As global supply chains become increasingly complex and expensive organizations are under growing pressure to automate and optimize their processes GreyMatter and gStore are harnessing the power of AI to perform more than a million optimizations every minute for customers across four continents increase revenue and deliver exceptional customer experiences,” said Akash Gupta “We invited enVista to join our Certified Partner Network based on its proven implementation methodology and decades of expertise as a trusted system integrator We look forward to supporting enVista and its clients with warehouse automation and store solutions powered by our software.”  The Roding Roadster is a bespoke carbon-fiber sportscar running a mid-engine BMW M twin-turbo six-cylinder  Its factory output of about 320 ponies is stocked higher via Daehler Competition  Output is now up to 450Hp — and put down to the pavement via a tweaked grip and suspension setup The looks are a bit hit and miss — but are a clear expression of ultra-low-volume carbon fiber production methods Roding Roadster R1 as exclusive Daehler competition line   Based upon long years of experience and competence in fields of optimized BMW models and power uprated BMW engines Daehler Design & Technik GmbH has found the matching answer for the absolutely exclusive Roding Roadster R1 as response to the specific needs of its clients in regard to a car with still more individual tuning The Roding Automobile GmbH Company at Roding in the Upper Palatinate (technologies company and manufacturer of the exclusive lightweight carbon fiber made sports car Roding Roadster in small-scale series founded in 2008) has invited the Daehler Design & Technik GmbH Company as development partner on board «Our Swiss partner Christoph Daehler brought along precious perceptions resulting out of long years of experience and practice and supporting us to realize the imaginations of our clients in a more innovative and individual way» which was the foundation for the Roding Roadster R1 in execution as «dÄHLer competition line» The absolutely exclusive car is matchless in the truest sense of the word corresponding to the highest demands concerning performance also in fields of environmental protection the Swiss confederate team is responsible for chassis balancing and the wheel-tire-combination The latter led to a complete modification of the now harmonious front and rear part of the vehicle The development process of the Roding Roadster R1 «dÄHLer  competition line» was influenced not only by modified design elements There appeared also a 6-piston high performance brake system optimized chassis and several optimizations of technical details which have been taken over also for the basic model Thanks to the experience and competence of Daehler Design & Technik GmbH gathered since many years in fields of power uprated BMW engines the Roding mid engine sports car with carbon fiber chassis and 340 HP (=250 kW)/450 Nm as basic performance is available in two power uprating stages (LS): the Daehler Design & Technik GmbH at Belp near Bern city stands for innovative techniques and precise highest level handicraft The foundation to survive so many years on a dynamic market with hard competition challenges in permanence are real visions sales flair and a good part of perseverance and endurance Christoph Daehler and his highly qualified team prove again that this is possible with success says owner Christoph Daehler: «That is craftsmanship accuracy and passion for detail – no matter The works have to be done with most of care – always: ‹made by dÄHLer›» The precious know-how for components of own production allows client’s service on highest level because the handling of high-tech products demands high-grade carefulness and expertise mobility is elevated to art form – more precisely: «the art of mobility» For more details concerning this indeed very creative cooperation please contact directly Tom Burkart is the founder and managing editor of Car-Revs-Daily.com an innovative and rapidly-expanding automotive news magazine He holds a Journalism JBA degree from the University of Wisconsin – Madison Burkart is available for all questions and concerns by email Tom(at)car-revs-daily.com He was also the originator of the poll tax policy that eventually brought the prime minister down in 1990 By that time, Jenkin, who has died aged 90, was already in the House of Lords but it was he who commissioned the original ministerial studies looking into reforming local government finance with a brief to replace rates on property with a tax that everyone would have to pay he played a large part in the Thatcher government’s early privatisation efforts but often contrived to give the impression of political haplessness and he was replaced by his more resolute and quick-footed deputy he urged the public to save electricity by brushing their teeth in the dark It then emerged that Jenkin himself used an electric toothbrush and his north London home was photographed with lights on in every room This off-the-cuff remark and the reaction to it perhaps pointed to why Jenkin although periodically described as a “big beast” of the Tories in the 1970s and 80s ultimately never quite reached the political heights The ministerial posts he held – chief secretary to the Treasury and energy minister under Heath then successively health and social services secretary industry secretary and environment secretary in Thatcher’s cabinets – were worthy rather than the chief offices of state to which he might have aspired Although he started off as a Keynesian Tory at the Treasury, he saw the way the wind was blowing and latterly followed the monetarist line. In his valedictory speech in the Lords in December 2014 when he became the first peer to resign on grounds of age he complained characteristically mildly that he had never made it to the front row in the annual cabinet photograph Patrick was the great-grandson of Henry Fleeming Jenkin, the inventor of the cable car and professor of engineering at Edinburgh University; the grandson of Charles Frewen Jenkin the first professor of engineering science at Oxford; and the son of Charles an industrial chemist for the Shell oil company Throughout his education at the Dragon school in Oxford he went on to chair various parliamentary science and technology committees and pronounced himself flabbergasted when given an award for making an outstanding contribution to scientific study by urging scientists to communicate more with the public went into industry as secretary of the Distillers Company’s chemicals and plastics division (1957-70) and served on Hornsey borough council (1960-63) In 1964 he succeeded Sir Winston Churchill as the Tory MP for Wanstead and Woodford in north-east London’s Essex suburbs, a seat he held until translated to the Lords in 1987. He was soon marked out as ministerial material and became an opposition Treasury spokesman under his mentor and hero Iain Macleod The latter’s sudden death within a few weeks of becoming chancellor of the exchequer following Heath’s election victory in 1970 was deeply destabilising to the new government and its economic policy Jenkin called it an appalling disaster and he promoted two years later to become chief secretary found himself having to implement public expenditure cuts to curb inflation the first inklings of what was to become standard Tory policy in coming decades His eventual promotion to energy minister came in the midst of the three-day week crisis just before the government went down to defeat in the February 1974 election Jenkin loyally supported Heath’s leadership bid against Thatcher but thereafter served as a frontbench spokesman for energy and then social security and health taking charge of that department in office following the Tories’ return to power in 1979 In 1981 he became industry secretary and two years later environment secretary supposedly a safe pair of hands as Thatcher sought first to transform the government’s relationship with the nationalised industries and then local government There was no sign that Jenkin dissented in any way from these policies and he initiated reforms starting with the process of decentralising the NHS and then linking state pension increases only to the rise in prices he was responsible for the privatisation of British Telecom His move in 1983 to the environment ministry to take on local government and particularly the inner-city Labour authorities was much less happy Radical Labour leaders such as Livingstone were undoubtedly provocative as they sought to challenge the government sometimes at the expense of running their own authorities efficiently But Thatcher was increasingly antipathetic to all council independence and Jenkin found himself in direct confrontation constantly wrong-footed by Livingstone over the Greater London council and having to explain why the administration was so hostile to local democracy It was an awkward stance that ultimately also harmed the Tories hamstringing their councillors and inadvertently cutting off the local government experience that many of the party’s own MPs had profited from before entering the Commons Ken Livingstone and Patrick Jenkin in 1985As it was charged with capping local authority spending and then with abolishing the metropolitan councils and the GLC altogether found support even on the government benches and in the Lords becoming ambivalent As a disgruntled Heath remarked: “The government had achieved the inconceivable in swinging the population of London behind Livingstone.” As a corollary – and perhaps more to the former Treasury minister’s taste – Jenkin commissioned an inquiry into the reform of the rating system which eventually came up with the poll tax inequity and sheer inoperability would bring down Thatcher Jenkin’s inability to deliver local government reform or present the reasons for it convincingly led to his sacking in 1985 and he stood down as an MP to go to the Lords at the general election two years later He resumed both a business career with directorships and advisory roles of which the most important was his chairmanship of Friends Provident (1988-98) and serving on public bodies such as his chairmanship of the Forest Healthcare NHS trust (1991-97) He was an assiduous attender and speaker in the Lords but taking advantage of a reform of the house’s procedures In his farewell speech he said: “I am getting on a bit … I have done what I can offer and it is best to bow out and let others carry on … there has to be a constant infusion of new blood with people who have current experience so it seems incumbent on oldies to hand over to a younger generation.” Despite his technocratic reputation, Jenkin had a liberal streak: early in his parliamentary career, he and a handful of other young Tories had supported the imposition of sanctions on Ian Smith’s breakaway regime in Rhodesia (Zimbabwe) and much later he became a convinced supporter of gay marriage Referring to his speech in the Lords in a debate on the issue in 2012, he told the Daily Telegraph: “I finished with a piece of theology which said that the love between two people has its parallels with the love of God for the human race” – and found himself inundated with messages of support from gay campaigners as a result Jenkin was an accomplished musician and gardener carpenter and bricklayer at the family’s holiday home in Scotland born 7 September 1926; died 20 December 2016 The collaboration between the two companies was possible thanks to Dahler’s years of experience in optimizing BMW engines, as the aforementioned sports car is powered by a Bavarian-built powerplant. Actually, it was purposefully designed by BMW to fit in the transverse layout of the Roding Roadster instead of the longitudinal like in its own models Daehler Design & Technik GmbH is offering the BMW N54 straight six-cylinder in two stages of tune: with 408 HP (402 hp) 540 Nm of torque and a maximum speed of 300 km/h (186 mph) 620 Nm of torque and a 310 km/h (192 mph) top speed just imagine that the vehicles weighs below 950 kg – which translates to a power to weight ratio below 3 kg/HP Featuring a lightweight construction with a carbon fiber reinforced plastic (CFRP) chassis cell that keeps curb weight at just 950kg (2,094 pounds) the Roadster draws its power from a mid-mounted 316hp (320PS) 3.0-liter turbocharged straight-six engine borrowed from BMW A six-speed manual transmission with a mechanical limited slip differential (TORSEN system with up to 40 percent locking ratio) feed the rear wheels with Roding quoting a zero to 100km/h (62mph) sprint time of just 3.9 seconds and a top speed of 285km/h (177mph) We’re also told that the Roadster can come to a complete stop from a speed of 100km/h (62mph) in 34.5 meters (113.2 feet) The model on display at the Swiss show is the first of the 23 special edition variants planned for sale by Roding Redbridge Council has been awarded £3m as part of the Greater London Authority’s (GLA) Civic Partnership Programme to transform the walking route between Wanstead Park delivering new green space and a stunning riverside path As one of five boroughs selected for the award Leader of Redbridge Council said “There’s so much to see and do across Redbridge and by improving pedestrian routes we’re connecting Wanstead Woodford and Ilford making sure people from all parts of the borough can access all we have to offer “Redbridge is a growing and changing borough and we’re working with partners including the GLA to make sure our growth works for everyone – creating more public spaces in addition to welcoming external investment new businesses and delivering leisure facilities.” authors explain how they are putting the concept of ‘wild service’ into practice A forthcoming book by a diverse band of right to roam campaigners offers a radical new vision of how people can repair both the natural world and their broken relationship to it Wild Service: Why Nature Needs You, inspired by the rare wild service tree calls on communities to develop new relationships with the natural world combining the hard graft of conservation science with the ceremony Indigenous traditions and even church services Jon Moses: ‘The idea of this book is to stitch the threads together.’ Photograph: Martin Godwin/The Guardian“All around the country we’re already seeing wild service – a huge flowering of grassroots ecology in the last 10 years,” says Jon Moses “But there hasn’t been a narrative binding all that energy The idea of this book is to stitch the threads together.” campaign against illegal sewage discharges and open up riverbank pathways His group recently spent a weekend clearing rubbish planting trees – including a wild service – and improving the riverside path But wild service, which one contributor suggests could become a voluntary national service, is not just hard labour for nature. As the folk singer singer Sam Lee writes in his chapter it encompasses paying homage through poetry and song sparking a new culture that will return wild species to the heart of human life “We know that people are disconnected from nature and we know that the restoration of nature is the work of this century,” says Powlesland I’ve found that hippy ceremonies are often ungrounded in action but equally a lot of action is ungrounded in ceremony Nature restoration days are sometimes a complete slog We’re trying to give people a bit more joy rather than just a muddy river and seven hours hard labour collecting rubbish.’ “If regarding nature as sacred happens in the UK it’s not going to come from the politicians saying we now believe in this’; it’s going to come from a grassroots movement of people who are connected with a specific local nature who then demand rights for nature on a national level.” For co-author Nick Hayes a renaissance of nature in culture can only flourish alongside challenging property rights that exclude others from accessing land Nick Hayes (left): ‘I’ve got no time for God – just its original source.’ Photograph: Martin Godwin/The Guardian“If you take away the notion of property as we’ve defined it in western law how do you enact belonging in Indigenous cultures From New Zealand to North America to Australia to India When communities come together on the land The only vestige that we’ve got [in England] is a church service – rituals and songs that everyone knows from childhood I’ve got no time for God – just its original source Hayes and Moses compare the church in the Berkshire village of Englefield whose doors are open every day with the thousands of acres of a nearby estate which “It’s only now that we have this idea that land ownership equates to absolute dominion and anyone forming their own relationship of care or connection to the land around them is prohibited,” says Moses one of only 3% of rivers in England and Wales with a statutory right of navigation and historic access to its banks Now it is home to a vociferous local campaign against river pollution “Those two things are obviously connected,” argues Moses “No one will describe what’s happening on the Wye as an access story but it absolutely is How do we protect people from themselves and protect nature from everyone?’ This innate sense of caution is completely blind to the magic that bringing people into the land has done for the environment.” Open churches suffer from theft and there is concern from conservationists and landowners that extending the right to roam will damage rare species with chicks of ground-nesting birds destroyed by an off-the-lead dog “Undoubtedly there will be repercussions because people don’t have a generational relationship with the Earth,” says Hayes But [opponents] will try and pin that on [the] right to roam when the source is people’s exclusion in the first place – landowners’ barbed wire fences.” Right to roam campaigner Nadia Shaikh writes a powerful chapter on the colonial and exclusionary mindset of mainstream conservation “I’ve worked in nature conservation for so many years and so many meetings are centred around really not liking people,” she says She thinks the sector suffers from a “toxic positivity” – unable to criticise itself – because practitioners are so attached to their self-image as “good” people doing “good” for nature Nadia Shaikh (right): ‘So many meetings are centred around really not liking people.’ Photograph: Martin Godwin/The GuardianConservation charities “say that people need to connect to nature in order to care for it and make change But then understanding what it is for everyone to connect to nature in a meaningful way Shaikh undertook community engagement for the RSPB when it assumed the management of Sherwood Forest ‘You do realise it’s the biggest collection of ancient oaks in Europe and these trees can’t be climbed on.’ But the tree-climbing creates relationships that in the future the charity will want to capitalise on – the children will become members but also the people who love and care for the place.” The book’s publication on 25 April is followed by book discussion groups and collaborative events with grassroots green collectives across the country “This is not the definitive conclusion of what wild service means,” says Hayes “It’s a provocation – throwing it open to other people to define it We’re just trying to say that belonging has an active dynamic to it.” People are often desperate to help but overwhelmed by the scale of global heating and ecological breakdown “We’ve been breastfed on leadership,” says Hayes “People have forgotten their own collective community power We’ve been taught that coming together is somehow seditious but it is the true power and the true way Wild Service by Nick Hayes (Bloomsbury, £20). To support the Guardian and Observer, order your copy at guardianbookshop.com Environment EditorTuesday February 28 2023 The TimesMore than three quarters of rivers tested in England breach proposed safety levels of “forever chemicals” and leading campaigners have warned of a “toxic timebomb” from the pollutants The worst river for the group of chemicals — per- and poly-fluoroalkyl substances (PFAS) — was the River Roding in East London which had concentrations more than 20 times higher than safety limits the Mersey in Cheshire and the Ouse in Bedfordshire were all found at least ten times over the limit PFAS are a group of chemicals used since the 1940s for their ability to repel oil and water in products such as non-stick pans Dubbed forever chemicals because of their persistent nature the authorities are concerned they may be linked to lower birth weights and higher cholesterol levels and other effects on health Analysis of Environment Agency data for 105 rivers has now revealed several hotspots of PFAS pollution in English rivers The research by the Rivers Trust revealed that 81 out of 105 rivers tested for the group of chemicals exceeded a proposed EU standard which is expected to become law later this year with 44 of 105 exceeding it by more than five times Those included the River Trent at Rugeley in Staffordshire the Arun near Horsham in West Sussex and the Yare in Norfolk Dr Rob Collins at the Rivers Trust said the levels found were shocking and warned they could be “just the tip of the iceberg” because monitoring of forever chemicals is so patchy who lives on a boat on the river in Barking said it was “very worrying” to hear the Roding was so contaminated “We do a lot of restoration work on the river such as if Thames Water puts sewage in,” he said “But these chemicals are really scary because what happens if we can’t get rid of them I swim in it in the summer and want to start a swimming club.” Powlesland added that the charity Thames 21 is hoping to win a bathing water designation for a beach-like area on the river near by Only two stretches of river have bathing water designation in England but The Times Clean It Up campaign is calling for hundreds more rivers to be designated Designation means greater monitoring but is not considered a silver bullet for cleaner water none of England’s rivers meet “good” status for chemical pollution which includes the PFAS group of chemicals The government does not expect the country’s water bodies to meet good chemical status until 2063 Campaigners said the new analysis showed it was vital to stop further chemical pollution chief executive of Wildlife and Countryside Link which commissioned the Rivers Trust research said: “No programme to clean up British rivers would be complete without a plan to prevent PFAS pollution which our research has found at levels which should ring alarm bells “The government should ban unnecessary forever-chemical use in products like cosmetics and food packaging and set safety standards to prevent dangerous chemical cocktail effects in the environment.” Government officials have previously suggested that forever chemicals could be “widely present” in English rivers and lakes The Rivers Trust took an average concentration of each individual PFAS chemical between 2019 and 2022 The group then compared the levels found with the proposed EU Environmental Quality Standard which the European parliament is considering before it goes to the European Commission with the expectation it will become law this year The UK has not put forward any proposals similar to the EU But it does have an upcoming Chemicals Strategy which Wildlife and Countryside Link want to see curbing PFAS use The government has launched a working group on PFAS and is planning to publish a review in spring on the risks from the chemicals A government spokesman said: “We are working at pace across government to assess the levels of PFAS occurring in the environment their sources and potential risks to inform future policy and regulatory approaches.” The Times is demanding faster action to improve the country’s waterways. Find out more about the Clean It Up campaign What’s the water like in your area?Is there a story that we need to cover The council was held as part of a Water Post 2043 project which the civil service ‘Policy Lab’ division has worked on with a team in DEFRA called DEFRA Futures The aim of the project is “to explore what decision-making in relation to the freshwater system could look like post 2043 The idea of an interspecies council is one developed by an organisation called Moral Imaginations, and uses “semi-improvisational, participatory techniques to bring the voice of nature into organisational decision-making”, according to a blog published by Policy Lab following the event The experimental council focused on the river Roding and took place in a location next to the river in Barking 24 participants including DEFRA and Policy Lab staff along with those with professional or community interest in the local area came together “to imagine and empathise with the needs of some of the species living in and around the river Roding”.  Each participant was given a different animal or part of the natural world - such as a river The group asked questions such as “What concerns does the bee have The blog said: “While we can’t truly know the answers to these questions the process of stepping out of our own shoes can help to deepen empathy and create new perspectives More-than-human thinking asks us to engage with the needs of both humans and other species in decision-making recognising that our actions often have an impact beyond people-centred considerations.”   Commenting in a video published alongside the blog said: “I must admit I probably as much as anyone else went into this with a bit of scepticism about [whether we could] just come out with the same insights if we weren’t thinking from the species perspective when we started getting into interspecies dialogue with one another it really brought new insight into the whole way we think about how decisions should be formed with those different ecological relationships.” The Policy Lab blog stated that the interspecies council approach “acted as a levelling tool bringing people with different roles and experiences together and allowing them to find common goals”.  The blog continued: “Rather than achieving an easy consensus the discussions amongst the group highlighted areas of tension which then prompted reflection about potential solutions and compromises we saw an appetite for people to keep engaging weeks after the council had taken place.  “Feedback also suggests that a legacy effect of more-than-human empathy has developed for some; almost all participants reported a noticeable lasting change within their perception or feelings towards nature the world or themselves in the week after the council.” The River Roding Interspecies Council was led by Moral Imagination’s founder Phoebe Tickell According to the organisation’s website it is “driving a movement of imagination-powered activism”.  strategies and movements [to] place life back at the centre of the economy and help catalyse futures that are rooted in values and a connection to a moral sense of what is important.” ‘Please enforce the law’: What you need to know about the river Wye legal challenge PopCons: Truss aims fire at 'unaccountable' EA and Natural England at launch of right-wing Tory group How a Land Use Framework could balance nature including legislation summaries to keep up to date with compliance deadlines Plan ahead with our Calendar feature highlighting upcoming compliance deadlines Planning Bill is a ‘regression’ in environmental law ‘Landmark moment’: UK wins right to keep sandeel fishing ban in UK-EU dispute Nature minister denies Planning Bill ‘repeals habitat protections’ DEFRA publishes legally required environment report after year-long delay Pressure mounts on UK and EU to ‘reset’ energy and environment partnership VW Sell More EVs, Yet Makes a Lot Less Money Ram Reveals Revamped, Stylish Sub-$50K 1500 Express Home/NewsRoding Roadster is fit for a role in The AvengersThe Roding Roadster 23 is here and having been faithful to its renderings We knew it would have proper fenders from the latest drawings but the red-and-black livery makes the little roofless coupe pop The Roding Roadster 23 is here, and having been faithful to its renderings Roding still hasn't decided to spill all of its secrets, not that it looks to have many, but we know there'll be just 23 examples offered. Behind those huggy seats is a turbocharged inline six-cylinder engine from BMW producing 320 horsepower and 332 pound-feet charged with propelling a car that weighs just 950 kilograms (2,090 pounds) An installable hood goes in the trunk section up front In the information from the company's site below Roding indicates that “the first cars to be produced” are a limited edition of 23 which makes it sound like they'll be around for more than a couple dozen cars With a design that shows character and temperament and uncompromisingly high performance this distinctive vehicle concept sets new standards in the roadster sports car segment It's supreme agility and powerful performance allow you to experience the art of German engineering – firsthand Because every Roding is manufactured by hand in a modern production process and can be customized to the smallest detail upon request each car becomes a very personal item for it's owner It is time to travel beyond the beaten path and experience true a Roadster created by people with passion for performance automobiles The Roding is a two seat mid-engine sports car with a carbon fiber chassis and a powerful BMW 6 cylinder turbo charged engine Maximum performance and driving dynamics are achieved through consequent lightweight construction and a state of the art six cylinder turbo charged engine The Roding Roadster is powered by a BMW high-efficiency six in line turbo engine The engine generates 320 HP and has a torque of 450 Newton meters The centre engine is installed in transverse configuration in relation to the vehicle longitudinal axis The lightweight concept with a vehicle weight of 950 kg in conjunction with the powerful engine give the Roding a power to weight ratio of less than 3 kg/HP This allows not only spectacular acceleration values but also high curve speeds which is expressed in a curve acceleration of 1.4 g Shifting is accomplished with a manual 6-Speed transmission which transfers the power to an limited slip differential with a locking effect of up to 40% This divides the engine power between the rear wheels increasing the available traction while accelerating In combination with a sporty mixed tire compound which can also optionally be ordered as a semi-slick variation the rear axle thereby always remains easy to control despite the high engine performance – even with the desired drift the shocks and the fully independent suspension with double wishbones can be adapted to the purpose – road or race track – and individually adjusted by the driver The first cars to be produced are a special edition hence the name “Roding Roadster 23”.These edition models feature an exclusive exterior design with all visible carbon parts like the hood doors and roof unpainted and the rest of the body in any custom paint color the customer wishes It also includes factory installed racing components This weight reduced Roadster was designed for the racetrack but is also completely suited for daily use The essence of lightweight construction is the concentration on only what is necessary and the application of modern materials The main chassis cell is manufactured of carbon fiber reinforced plastic (CFRP) while the front and rear subframes are aluminum-CFRP hybrid structures This material has been proven in its application in aerospace and formula one due to its low weight and high stiffness Characteristics that play a large role in an open roadster like the Roding since stiffness and light weight are the foundation for precise handling and impressive road performance Another core element of the main chassis is the integration of the required functions. For instance, the seat mounting points are integrated in the form making adapter brackets unnecessary “The Roding shows how it is possible to make ideal use of light-weight potentials in the vehicle with the fiber bond-appropriate design The technologies used clear the way for the structural use of fiber bond materials in the major series” one of the most renowned scientists and technology specialists in the area of CFRP 2-seater roadster with mid engine and rear wheel drive Convertible and coupé through removable roof shell which can be stored in the vehicle One-piece carbon fiber passenger compartment with integrated roll-over structure Front and rear section with carbon-fiber-aluminum hybrid design Front trunk with storage area for the roof shells and the ability to load long objects such as a pair of skis LLC and respective content providers on this website Other product and company names shown may be trademarks of their respective owners LLC and respective content providers to this website may receive compensation for some links to products and services on this website Two manufacturers of sports cars hailing from Germany are planning to debut some new models at the 2012 Geneva Motor Show taking place next month Follow Motor Authority on Facebook, Twitter, and Google+ We’ll forgive you if you’ve never heard of the name Roding before the company made its first appearance at the 2009 Frankfurt Motor Show with an odd-looking roadster concept promising to deliver a production version of the two-seat mid-engine sports car in the near future Roding says that the main chassis cell is made of carbon fiber reinforced plastic (CFRP) while the front and rear subframes are aluminum-CFRP hybrid structures The small roadster model’s body is made of fiberglass sandwich panels (CFRP optional) to further reduce the overall weight The Roadster is being tested with what the company describes as a “high-efficiency six-cylinder turbo engine coming from a car manufacturer with reputation” delivering over 300-horses and more than 400Nm (295 lb-ft) of peak torque Given the specs of the engine and the fact that Roding is based in Germany chances are the engine is sourced from either BMW or Audi the production Roadster will make its debut in the spring of 2012 Here at the Geneva Motor Show 2012 you get all sorts of manufacturers Roding are a relative newcomer and this year they have unveiled the bespoke Roding Roadster 23 a limited production sports car with a lightweight A lot of work has been pumped into this project and we suspect the results will probably speak for themselves in performance terms Each will be a mid-engine sports car with a carbon fiber chassis and a BMW in-line 3.0 liter six-cylinder turbocharged engine Each will develop a peak output of 320 horsepower at 5,800rpm and 450Nm of torque between 1,300 – 4,500rpm The key ingredient will be the 950kg target weight This makes 0-100km/h possible in 3.9 seconds with a top speed of 285km/h Clearly this project conforms to the old school principles of modest power and website in this browser for the next time I comment You will also start receiving the Star's free morning newsletter Local women held a “solidarity walk” following recent sexual assaults in Roding Park The eight hectare North York park has a baseball diamond as well as walking trails and connects a TCHC apartment building with a community centre at the north end The park and the surrounding area fall into Downsview-Roding-CFB one of Toronto’s 31 neighbourhood improvement areas identified as at risk and needing special attention from the city based on 15 indicators such as unemployment rate and voter turnout Just over half of the people in the neighbourhood speak a language other than English or French as their mother tongue The community has comparatively good access to green space but not a good overall walk score shown behind neighbourhood women on a ladies solidarity walk in response to recent sexual assaults offers recreational programs like adult volleyball and children’s guitar classes as well as a skating rink and outdoor pool City councilor wants to get more eyes into the park by putting in more exercise equipment Maria Negri squints in the spring late-day sun pointing to a rolling hill dotted with dandelions where she used to sit with her brother to watch baseball She has lived around the corner from sprawling Roding Park for 39 years The park is an oasis in the middle of an ordinary suburban neighbourhood a rare green space for the eight storey TCHC building that overlooks it bordered on the other side by a thriving community centre that offers yoga and seniors’ Tai Chi But lately it’s been drawing headlines for sexual assault There have been three reported in the park since 2014 “Something has to be done to make it safe and so we can enjoy it,” said Negri one of about 15 neighbourhood women wearing light green ribbons on a ladies solidarity walk through the park on a recent weekday evening “The last one really made me nervous because it was noon Toronto Police put out a sketch of a suspect later that month Spokesperson Victor Kwong said the investigation is ongoing and there’s no more information to provide at this time Not long ago the community was billed as a TCHC success story heralded as a triumph in mixed-use housing the apartments at Roding Park Place include 98 subsidized units and 26 at market rent Local city councillor Maria Augimeri said she wants to get more people into the park which sits tucked back from the street with some areas hidden from sight Some people have suggested women shouldn’t be walking in the park alone but that’s something Augimeri doesn’t accept “If a man can walk in Roding Park a woman should be able to walk in Roding Park,” she said Augimeri will be applying for grants to put in adult exercise equipment and will ask council in the next few months for money to expand the existing children’s playground She also plans to ask staff to look into putting cameras into the park Women are twice as likely to be sexually assaulted by a man they know than by a stranger but the assaults in the park have left many neighbours on edge a tenant representative at the nearby Roding Park Place said he hasn’t let his kids walk alone there since hearing about it “Since that incident happened it’s like 50 per cent less people walking through alone” he said Oliva who has lived at Roding Park Place for about fifteen years building is cleaner and safer than other TCHC communities the elevators generally work and people are more engaged with the community The problems — like a recent break in to a car in the underground parking lot — are caused by people who don’t live in the community given recent events he would like to see TCHC special constables patrol more often who has lived in the neigbourhood for 14 years said the community needs to “take back” the park and she’s noticed people not using it as much after the most recent sexual assault beautiful community,” with many seniors and a prominent Italian population facing the same challenges as other neighborhoods across the city no one would even think twice about walking alone in Roding Park Your browser is out of date and potentially vulnerable to security risks.We recommend switching to one of the following browsers: Account processing issue - the email address may already exist Invalid password or account does not exist Submitting this form below will send a message to your email with a link to change your password An email message containing instructions on how to reset your password has been sent to the email address listed on your account